API Reference

This page contains a comprehensive list of all classes and functions within lhotse.

Datasets

PyTorch Dataset wrappers for common tasks.

Speech Recognition

class lhotse.dataset.speech_recognition.SpeechRecognitionDataset(cuts)

The PyTorch Dataset for the speech recognition task. Each item in this dataset is a dict of:

{
    'features': (T x F) tensor,
    'text': string,
    'supervisions_mask': (T) tensor
}

The supervisions_mask field is a mask that specifies which frames are covered by a supervision by assigning a value of 1 (in this case: segments with transcribed speech contents), and which are not by asigning a value of 0 (in this case: padding, contextual noise, or in general the acoustic context without transcription).

In the future, will be extended by graph supervisions.

__init__(cuts)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.dataset.speech_recognition.K2SpeechRecognitionIterableDataset(cuts, max_frames=26000, max_cuts=None, shuffle=False, concat_cuts=True, concat_cuts_gap=1.0, concat_cuts_duration_factor=1)

The PyTorch Dataset for the speech recognition task using K2 library.

This dataset internally batches and collates the Cuts and should be used with PyTorch DataLoader with argument batch_size=None to work properly. The batch size is determined automatically to satisfy the constraints of max_frames and max_cuts.

This dataset will automatically partition itself when used with a multiprocessing DataLoader (i.e. the same cut will not appear twice in the same epoch).

By default, we “pack” the batches to minimize the amount of padding - we achieve that by concatenating the cuts’ feature matrices with a small amount of silence (padding) in between.

Each item in this dataset is a dict of:

{
    'features': float tensor of shape (B, T, F)
    'supervisions': [
        {
            'cut_id': List[str] of len S
            'sequence_idx': Tensor[int] of shape (S,)
            'text': List[str] of len S
            'start_frame': Tensor[int] of shape (S,)
            'num_frames': Tensor[int] of shape (S,)
        }
    ]
}

Dimension symbols legend: * B - batch size (number of Cuts) * S - number of supervision segments (greater or equal to B, as each Cut may have multiple supervisions) * T - number of frames of the longest Cut * F - number of features

The ‘sequence_idx’ field is the index of the Cut used to create the example in the Dataset.

__init__(cuts, max_frames=26000, max_cuts=None, shuffle=False, concat_cuts=True, concat_cuts_gap=1.0, concat_cuts_duration_factor=1)

K2 ASR IterableDataset constructor.

Parameters

• cuts (CutSet) – the CutSet to sample data from.

• max_frames (int) – The maximum number of feature frames that we’re going to put in a single batch. The padding frames do not contribute to that limit, since we pack the batch by default to minimze the amount of padding.

• max_cuts (Optional[int]) – The maximum number of cuts sampled to form a mini-batch. By default, this constraint is off.

• shuffle (bool) – When True, the cuts will be shuffled at the start of iteration. Convenient when mini-batch loop is inside an outer epoch-level loop, e.g.: for epoch in range(10): for batch in dataset: … as every epoch will see a different cuts order.

• concat_cuts (bool) – When True, we will concatenate the cuts to minimize the total amount of padding; e.g. instead of creating a batch with 40 examples, we will merge some of the examples together adding some silence between them to avoid a large number of padding frames that waste the computation. Enabled by default.

• concat_cuts_gap (float) – The duration of silence in seconds that is inserted between the cuts; it’s goal is to let the model “know” that there are separate utterances in a single example.

• concat_cuts_duration_factor (float) – Determines the maximum duration of the concatenated cuts; by default it’s 1, setting the limit at the duration of the longest cut in the batch.

lhotse.dataset.speech_recognition.concat_cuts(cuts, gap=1.0, max_duration=None)

We’re going to concatenate the cuts to minimize the amount of total padding frames used. This is actually solving a knapsack problem. In this initial implementation we’re using a greedy approach: going from the back (i.e. the shortest cuts) we’ll try to concat them to the longest cut that still has some “space” at the end.

Parameters

• cuts (List[Union[ForwardRef, ForwardRef, ForwardRef]]) – a list of cuts to pack.

• gap (float) – the duration of silence inserted between concatenated cuts.

• max_duration (Optional[float]) – the maximum duration for the concatenated cuts (by default set to the duration of the first cut).

:return a list of packed cuts.

Return type

List[Union[ForwardRef, ForwardRef, ForwardRef]]

class lhotse.dataset.speech_recognition.K2SpeechRecognitionDataset(cuts)

The PyTorch Dataset for the speech recognition task using K2 library. Each item in this dataset is a dict of:

{
    'features': (T x F) tensor,
    'supervisions': List[Dict] -> [
        {
            'sequence_idx': int
            'text': string,
            'start_frame': int,
            'num_frames': int
        } (multiplied N times, for each of the N supervisions present in the Cut)
    ]
}

The ‘sequence_idx’ field is the index of the Cut used to create the example in the Dataset. It is mapped to the batch index later in the DataLoader.

__init__(cuts)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.dataset.speech_recognition.K2DataLoader(*args, **kwds)

A PyTorch DataLoader that has a custom collate_fn that complements the K2SpeechRecognitionDataset.

The ‘features’ tensor is collated in a standard way to return a tensor of shape (B, T, F).

The ‘supervisions’ dict contains the same fields as in K2SpeechRecognitionDataset, except that each sub-field (like ‘start_frame’) is a 1D PyTorch tensor with shape (B,). The ‘text’ sub-field is an exception - it’s a list of strings with length equal to batch size.

The ‘sequence_idx’ sub-field in ‘supervisions’, which originally points to index of the example in the Dataset, is remapped to the index of the corresponding features matrix in the collated ‘features’. Multiple supervisions coming from the same cut will share the same ‘sequence_idx’.

For an example, see test/dataset/test_speech_recognition_dataset.py::test_k2_dataloader().

__init__(*args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

dataset

batch_size

num_workers

pin_memory

drop_last

timeout

sampler

prefetch_factor

lhotse.dataset.speech_recognition.multi_supervision_collate_fn(batch)

Custom collate_fn for K2SpeechRecognitionDataset.

It merges the items provided by K2SpeechRecognitionDataset into the following structure:

{
    'features': float tensor of shape (B, T, F)
    'supervisions': [
        {
            'sequence_idx': Tensor[int] of shape (S,)
            'text': List[str] of len S
            'start_frame': Tensor[int] of shape (S,)
            'num_frames': Tensor[int] of shape (S,)
        }
    ]
}

Dimension symbols legend: * B - batch size (number of Cuts), * S - number of supervision segments (greater or equal to B, as each Cut may have multiple supervisions), * T - number of frames of the longest Cut * F - number of features

Return type

Dict

Source Separation

class lhotse.dataset.source_separation.SourceSeparationDataset(sources_set, mixtures_set)

An abstract base class, implementing PyTorch Dataset for the source separation task. It’s created from two CutSets - one provides the audio cuts for the sources, and the other one the audio cuts for the signal mix. When queried for data samples, it returns a dict of:

{
    'sources': (N x T x F) tensor,
    'mixture': (T x F) tensor,
    'real_mask': (N x T x F) tensor,
    'binary_mask': (T x F) tensor
}

__init__(sources_set, mixtures_set)

Initialize self. See help(type(self)) for accurate signature.

validate()

class lhotse.dataset.source_separation.DynamicallyMixedSourceSeparationDataset(sources_set, mixtures_set, nonsources_set=None)

A PyTorch Dataset for the source separation task. It’s created from a number of CutSets:

• sources_set: provides the audio cuts for the sources that (the targets of source separation),

• mixtures_set: provides the audio cuts for the signal mix (the input of source separation),

• nonsources_set: (optional) provides the audio cuts for other signals that are in the mix, but are not the targets of source separation. Useful for adding noise.

When queried for data samples, it returns a dict of:

{
    'sources': (N x T x F) tensor,
    'mixture': (T x F) tensor,
    'real_mask': (N x T x F) tensor,
    'binary_mask': (T x F) tensor
}

This Dataset performs on-the-fly feature-domain mixing of the sources. It expects the mixtures_set to contain MixedCuts, so that it knows which Cuts should be mixed together.

__init__(sources_set, mixtures_set, nonsources_set=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.dataset.source_separation.PreMixedSourceSeparationDataset(sources_set, mixtures_set)

A PyTorch Dataset for the source separation task. It’s created from two CutSets - one provides the audio cuts for the sources, and the other one the audio cuts for the signal mix. When queried for data samples, it returns a dict of:

{
    'sources': (N x T x F) tensor,
    'mixture': (T x F) tensor,
    'real_mask': (N x T x F) tensor,
    'binary_mask': (T x F) tensor
}

It expects both CutSets to return regular Cuts, meaning that the signals were mixed in the time domain. In contrast to DynamicallyMixedSourceSeparationDataset, no on-the-fly feature-domain-mixing is performed.

__init__(sources_set, mixtures_set)

Initialize self. See help(type(self)) for accurate signature.

Unsupervised

class lhotse.dataset.unsupervised.UnsupervisedDataset(cuts)

Dataset that contains no supervision - it only provides the features extracted from recordings. The returned features are a torch.Tensor of shape (T x F), where T is the number of frames, and F is the feature dimension.

__init__(cuts)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.dataset.unsupervised.UnsupervisedWaveformDataset(cuts)

A variant of UnsupervisedDataset that provides waveform samples instead of features. The output is a tensor of shape (C, T), with C being the number of channels and T the number of audio samples. In this implemenation, there will always be a single channel.

class lhotse.dataset.unsupervised.DynamicUnsupervisedDataset(feature_extractor, cuts, augment_fn=None)

An example dataset that shows how to use on-the-fly feature extraction in Lhotse. It accepts two additional inputs - a FeatureExtractor and an optional WavAugmenter for time-domain data augmentation.. The output is approximately the same as that of the UnsupervisedDataset - there might be slight differences for MixedCut``s, because this dataset mixes them in the time domain, and ``UnsupervisedDataset does that in the feature domain. Cuts that are not mixed will yield identical results in both dataset classes.

__init__(feature_extractor, cuts, augment_fn=None)

Initialize self. See help(type(self)) for accurate signature.

Voice Activity Detection

class lhotse.dataset.vad.VadDataset(cuts)

The PyTorch Dataset for the voice activity detection task. Each item in this dataset is a dict of:

{
    'features': (T x F) tensor
    'is_voice': (T x 1) tensor
}

__init__(cuts)

Initialize self. See help(type(self)) for accurate signature.

Diarization (experimental)

class lhotse.dataset.diarization.DiarizationDataset(cuts, min_speaker_dim=None, global_speaker_ids=False)

A PyTorch Dataset for the speaker diarization task. Our assumptions about speaker diarization are the following:

• we assume a single channel input (for now), which could be either a true mono signal

  or a beamforming result from a microphone array.

• we assume that the supervision used for model training is a speech activity matrix, with one

  row dedicated to each speaker (either in the current cut or the whole dataset, depending on the settings). The columns correspond to feature frames. Each row is effectively a Voice Activity Detection supervision for a single speaker. This setup is somewhat inspired by the TS-VAD paper: https://arxiv.org/abs/2005.07272

Each item in this dataset is a dict of:

{
    'features': (T x F) tensor
    'speaker_activity': (num_speaker x T) tensor
}

Constructor arguments:

Parameters

• cuts (CutSet) – a CutSet used to create the dataset object.

• min_speaker_dim (Optional[int]) – optional int, when specified it will enforce that the matrix shape is at least that value (useful for datasets like CHiME 6 where the number of speakers is always 4, but some cuts might have less speakers than that).

• global_speaker_ids (bool) – a bool, indicates whether the same speaker should always retain the same row index in the speaker activity matrix (useful for speaker-dependent systems)

• root_dir – a prefix path to be attached to the feature files paths.

__init__(cuts, min_speaker_dim=None, global_speaker_ids=False)

Initialize self. See help(type(self)) for accurate signature.

Recording manifests

Data structures used for describing audio recordings in a dataset.

class lhotse.audio.AudioSource(type: str, channels: List[int], source: str)

AudioSource represents audio data that can be retrieved from somewhere. Supported sources of audio are currently: - ‘file’ (formats supported by librosa, possibly multi-channel) - ‘command’ [unix pipe] (must be WAVE, possibly multi-channel)

type: str

channels: List[int]

source: str

load_audio(offset_seconds=0.0, duration_seconds=None)

Load the AudioSource (both files and commands) with librosa, accounting for many audio formats and multi-channel inputs. Returns numpy array with shapes: (n_samples) for single-channel, (n_channels, n_samples) for multi-channel.

Return type

ndarray

with_path_prefix(path)

Return type

AudioSource

static from_dict(data)

Return type

AudioSource

__init__(type, channels, source)

Initialize self. See help(type(self)) for accurate signature.

lhotse.audio.read_audio(path, offset, duration)

Return type

Tuple[ndarray, int]

class lhotse.audio.Recording(id: str, sources: List[lhotse.audio.AudioSource], sampling_rate: int, num_samples: int, duration: float)

Recording represents an AudioSource along with some metadata.

id: str

sources: List[AudioSource]

sampling_rate: int

num_samples: int

duration: Seconds

static from_sphere(sph_path, recording_id=None, relative_path_depth=None)

Read a SPHERE file’s header and create the corresponding Recording.

Parameters

• sph_path (Union[Path, str]) – Path to the sphere (.sph) file.

• recording_id (Optional[str]) – recording id, when not specified ream the filename’s stem (“x.wav” -> “x”).

• relative_path_depth (Optional[int]) – optional int specifying how many last parts of the file path should be retained in the AudioSource. By default writes the path as is.

Return type

Recording

Returns

a new Recording instance pointing to the sphere file.

static from_wav(path, recording_id=None)

Read a WAVE file’s header and create the corresponding Recording.

Parameters

• path (Union[Path, str]) – Path to the WAVE (.wav) file.

• recording_id (Optional[str]) – recording id, when not specified ream the filename’s stem (“x.wav” -> “x”).

Return type

Recording

Returns

a new Recording instance pointing to the sphere file.

property num_channels

property channel_ids

load_audio(channels=None, offset_seconds=0.0, duration_seconds=None)

Return type

ndarray

with_path_prefix(path)

Return type

Recording

static from_dict(data)

Return type

Recording

__init__(id, sources, sampling_rate, num_samples, duration)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.audio.RecordingSet(*args, **kwds)

RecordingSet represents a dataset of recordings. It does not contain any annotation - just the information needed to retrieve a recording (possibly multi-channel, from files or from shell commands and pipes) and some metadata for each of them.

It also supports (de)serialization to/from YAML and takes care of mapping between rich Python classes and YAML primitives during conversion.

recordings: Dict[str, Recording]

static from_recordings(recordings)

Return type

RecordingSet

static from_dicts(data)

Return type

RecordingSet

to_dicts()

Return type

List[dict]

filter(predicate)

Return a new RecordingSet with the Recordings that satisfy the predicate.

Parameters

predicate (Callable[[Recording], bool]) – a function that takes a recording as an argument and returns bool.

Return type

RecordingSet

Returns

a filtered RecordingSet.

split(num_splits, randomize=False)

Split the RecordingSet into num_splits pieces of equal size.

Parameters

• num_splits (int) – Requested number of splits.

• randomize (bool) – Optionally randomize the recordings order first.

Return type

List[RecordingSet]

Returns

A list of RecordingSet pieces.

load_audio(recording_id, channels=None, offset_seconds=0.0, duration_seconds=None)

Return type

ndarray

with_path_prefix(path)

Return type

RecordingSet

num_channels(recording_id)

Return type

int

sampling_rate(recording_id)

Return type

int

num_samples(recording_id)

Return type

int

duration(recording_id)

Return type

float

__init__(recordings)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.audio.AudioMixer(base_audio, sampling_rate)

Utility class to mix multiple waveforms into a single one. It should be instantiated separately for each mixing session (i.e. each MixedCut will create a separate AudioMixer to mix its tracks). It is initialized with a numpy array of audio samples (typically float32 in [-1, 1] range) that represents the “reference” signal for the mix. Other signals can be mixed to it with different time offsets and SNRs using the add_to_mix method. The time offset is relative to the start of the reference signal (only positive values are supported). The SNR is relative to the energy of the signal used to initialize the AudioMixer.

__init__(base_audio, sampling_rate)

Parameters

• base_audio (ndarray) – A numpy array with the audio samples for the base signal (all the other signals will be mixed to it).

• sampling_rate (int) – Sampling rate of the audio.

property unmixed_audio

Return a numpy ndarray with the shape (num_tracks, num_samples), where each track is zero padded and scaled adequately to the offsets and SNR used in add_to_mix call.

Return type

ndarray

property mixed_audio

Return a numpy ndarray with the shape (1, num_samples) - a mono mix of the tracks supplied with add_to_mix calls.

Return type

ndarray

add_to_mix(audio, snr=None, offset=0.0)

Add audio (only support mono-channel) of a new track into the mix. :type audio: ndarray :param audio: An array of audio samples to be mixed in. :type snr: Optional[float] :param snr: Signal-to-noise ratio, assuming audio represents noise (positive SNR - lower audio energy, negative SNR - higher audio energy) :type offset: float :param offset: How many seconds to shift audio in time. For mixing, the signal will be padded before the start with low energy values. :return:

lhotse.audio.audio_energy(audio)

Return type

float

Supervision manifests

Data structures used for describing supervisions in a dataset.

class lhotse.supervision.SupervisionSegment(id: str, recording_id: str, start: float, duration: float, channel: int = 0, text: Union[str, NoneType] = None, language: Union[str, NoneType] = None, speaker: Union[str, NoneType] = None, gender: Union[str, NoneType] = None, custom: Union[Dict[str, Any], NoneType] = None)

id: str

recording_id: str

start: Seconds

duration: Seconds

channel: int = 0

text: Optional[str] = None

language: Optional[str] = None

speaker: Optional[str] = None

gender: Optional[str] = None

custom: Optional[Dict[str, Any]] = None

property end

Return type

float

with_offset(offset)

Return an identical SupervisionSegment, but with the offset added to the start field.

Return type

SupervisionSegment

trim(end)

Return an identical SupervisionSegment, but ensure that self.start is not negative (in which case it’s set to 0) and self.end does not exceed the end parameter.

This method is useful for ensuring that the supervision does not exceed a cut’s bounds, in which case pass cut.duration as the end argument, since supervision times are relative to the cut.

Return type

SupervisionSegment

map(transform_fn)

Return a copy of the current segment, transformed with transform_fn.

Parameters

transform_fn (Callable[[SupervisionSegment], SupervisionSegment]) – a function that takes a segment as input, transforms it and returns a new segment.

Return type

SupervisionSegment

Returns

a modified SupervisionSegment.

transform_text(transform_fn)

Return a copy of the current segment with transformed text field. Useful for text normalization, phonetic transcription, etc.

Parameters

transform_fn (Callable[[str], str]) – a function that accepts a string and returns a string.

Return type

SupervisionSegment

Returns

a SupervisionSegment with adjusted text.

static from_dict(data)

Return type

SupervisionSegment

__init__(id, recording_id, start, duration, channel=0, text=None, language=None, speaker=None, gender=None, custom=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.supervision.SupervisionSet(*args, **kwds)

SupervisionSet represents a collection of segments containing some supervision information. The only required fields are the ID of the segment, ID of the corresponding recording, and the start and duration of the segment in seconds. All other fields, such as text, language or speaker, are deliberately optional to support a wide range of tasks, as well as adding more supervision types in the future, while retaining backwards compatibility.

segments: Dict[str, SupervisionSegment]

static from_segments(segments)

Return type

SupervisionSet

static from_dicts(data)

Return type

SupervisionSet

to_dicts()

Return type

List[dict]

split(num_splits, randomize=False)

Split the SupervisionSet into num_splits pieces of equal size.

Parameters

• num_splits (int) – Requested number of splits.

• randomize (bool) – Optionally randomize the supervisions order first.

Return type

List[SupervisionSet]

Returns

A list of SupervisionSet pieces.

filter(predicate)

Return a new SupervisionSet with the SupervisionSegments that satisfy the predicate.

Parameters

predicate (Callable[[SupervisionSegment], bool]) – a function that takes a supervision as an argument and returns bool.

Return type

SupervisionSet

Returns

a filtered SupervisionSet.

map(transform_fn)

Map a transform_fn to the SupervisionSegments and return a new SupervisionSet.

Parameters

transform_fn (Callable[[SupervisionSegment], SupervisionSegment]) – a function that modifies a supervision as an argument.

Return type

SupervisionSet

Returns

a new SupervisionSet with modified segments.

transform_text(transform_fn)

Return a copy of the current SupervisionSet with the segments having a transformed text field. Useful for text normalization, phonetic transcription, etc.

Parameters

transform_fn (Callable[[str], str]) – a function that accepts a string and returns a string.

Return type

SupervisionSet

Returns

a SupervisionSet with adjusted text.

find(recording_id, channel=None, start_after=0, end_before=None, adjust_offset=False)

Return an iterable of segments that match the provided recording_id.

Parameters

• recording_id (str) – Desired recording ID.

• channel (Optional[int]) – When specified, return supervisions in that channel - otherwise, in all channels.

• start_after (float) – When specified, return segments that start after the given value.

• end_before (Optional[float]) – When specified, return segments that end before the given value.

• adjust_offset (bool) – When true, return segments as if the recordings had started at start_after. This is useful for creating Cuts. Fom a user perspective, when dealing with a Cut, it is no longer helpful to know when the supervisions starts in a recording - instead, it’s useful to know when the supervision starts relative to the start of the Cut. In the anticipated use-case, start_after and end_before would be the beginning and end of a cut; this option converts the times to be relative to the start of the cut.

Return type

Iterable[SupervisionSegment]

Returns

An iterator over supervision segments satisfying all criteria.

__init__(segments, _segments_by_recording_id=None)

Initialize self. See help(type(self)) for accurate signature.

Feature extraction and manifests

Data structures and tools used for feature extraction and description.

Features API - extractor and manifests

class lhotse.features.base.FeatureExtractor(config=None)

The base class for all feature extractors in Lhotse. It is initialized with a config object, specific to a particular feature extraction method. The config is expected to be a dataclass so that it can be easily serialized.

All derived feature extractors must implement at least the following:

• a name class attribute (how are these features called, e.g. ‘mfcc’)

• a config_type class attribute that points to the configuration dataclass type

• the extract method,

• the frame_shift property.

Feature extractors that support feature-domain mixing should additionally specify two static methods:

• compute_energy, and

• mix.

By itself, the FeatureExtractor offers the following high-level methods that are not intended for overriding:

• extract_from_samples_and_store

• extract_from_recording_and_store

These methods run a larger feature extraction pipeline that involves data augmentation and disk storage.

name = None

config_type = None

__init__(config=None)

Initialize self. See help(type(self)) for accurate signature.

abstract extract(samples, sampling_rate)

Defines how to extract features using a numpy ndarray of audio samples and the sampling rate.

Return type

ndarray

Returns

a numpy ndarray representing the feature matrix.

abstract property frame_shift

Return type

float

abstract feature_dim(sampling_rate)

Return type

int

static mix(features_a, features_b, energy_scaling_factor_b)

Perform feature-domain mix of two singals, a and b, and return the mixed signal.

Parameters

• features_a (ndarray) – Left-hand side (reference) signal.

• features_b (ndarray) – Right-hand side (mixed-in) signal.

• energy_scaling_factor_b (float) – A scaling factor for features_b energy. It is used to achieve a specific SNR. E.g. to mix with an SNR of 10dB when both features_a and features_b energies are 100, the features_b signal energy needs to be scaled by 0.1. Since different features (e.g. spectrogram, fbank, MFCC) require different combination of transformations (e.g. exp, log, sqrt, pow) to allow mixing of two signals, the exact place where to apply energy_scaling_factor_b to the signal is determined by the implementer.

Return type

ndarray

Returns

A mixed feature matrix.

static compute_energy(features)

Compute the total energy of a feature matrix. How the energy is computed depends on a particular type of features. It is expected that when implemented, compute_energy will never return zero.

Parameters

features (ndarray) – A feature matrix.

Return type

float

Returns

A positive float value of the signal energy.

extract_from_samples_and_store(samples, storage, sampling_rate, offset=0, augment_fn=None)

Extract the features from an array of audio samples in a full pipeline:

• optional audio augmentation;

• extract the features;

• save them to disk in a specified directory;

• return a Features object with a description of the extracted features.

Note, unlike in extract_from_recording_and_store, the returned Features object might not be suitable to store in a FeatureSet, as it does not reference any particular Recording. Instead, this method is useful when extracting features from cuts - especially MixedCut instances, which may be created from multiple recordings and channels.

Parameters

• samples (ndarray) – a numpy ndarray with the audio samples.

• sampling_rate (int) – integer sampling rate of samples.

• storage (FeaturesWriter) – a FeaturesWriter object that will handle storing the feature matrices.

• offset (float) – an offset in seconds for where to start reading the recording - when used for Cut feature extraction, must be equal to Cut.start.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – an optional WavAugmenter instance to modify the waveform before feature extraction.

Returns

a Features manifest item for the extracted feature matrix (it is not written to disk).

extract_from_recording_and_store(recording, storage, offset=0, duration=None, channels=None, augment_fn=None)

Extract the features from a Recording in a full pipeline:

• load audio from disk;

• optionally, perform audio augmentation;

• extract the features;

• save them to disk in a specified directory;

• return a Features object with a description of the extracted features and the source data used.

Parameters

• recording (Recording) – a Recording that specifies what’s the input audio.

• storage (FeaturesWriter) – a FeaturesWriter object that will handle storing the feature matrices.

• offset (float) – an optional offset in seconds for where to start reading the recording.

• duration (Optional[float]) – an optional duration specifying how much audio to load from the recording.

• channels (Union[int, List[int], None]) – an optional int or list of ints, specifying the channels; by default, all channels will be used.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – an optional WavAugmenter instance to modify the waveform before feature extraction.

Returns

a Features manifest item for the extracted feature matrix.

classmethod from_dict(data)

Return type

FeatureExtractor

classmethod from_yaml(path)

Return type

FeatureExtractor

to_yaml(path)

lhotse.features.base.get_extractor_type(name)

Return the feature extractor type corresponding to the given name.

Parameters

name (str) – specifies which feature extractor should be used.

Return type

Type

Returns

A feature extractors type.

lhotse.features.base.create_default_feature_extractor(name)

Create a feature extractor object with a default configuration.

Parameters

name (str) – specifies which feature extractor should be used.

Return type

Optional[FeatureExtractor]

Returns

A new feature extractor instance.

lhotse.features.base.register_extractor(cls)

This decorator is used to register feature extractor classes in Lhotse so they can be easily created just by knowing their name.

An example of usage:

@register_extractor class MyFeatureExtractor: …

Parameters

cls – A type (class) that is being registered.

Returns

Registered type.

class lhotse.features.base.TorchaudioFeatureExtractor(config=None)

Common abstract base class for all torchaudio based feature extractors.

feature_fn = None

extract(samples, sampling_rate)

Defines how to extract features using a numpy ndarray of audio samples and the sampling rate.

Return type

ndarray

Returns

a numpy ndarray representing the feature matrix.

property frame_shift

Return type

float

class lhotse.features.base.Features(type: str, num_frames: int, num_features: int, sampling_rate: int, start: float, duration: float, storage_type: str, storage_path: str, storage_key: str, recording_id: Optional[str] = None, channels: Optional[Union[int, List[int]]] = None)

Represents features extracted for some particular time range in a given recording and channel. It contains metadata about how it’s stored: storage_type describes “how to read it”, for now it supports numpy arrays serialized with np.save, as well as arrays compressed with lilcom; storage_path is the path to the file on the local filesystem.

type: str

num_frames: int

num_features: int

sampling_rate: int

start: Seconds

duration: Seconds

storage_type: str

storage_path: str

storage_key: str

recording_id: Optional[str] = None

channels: Optional[Union[int, List[int]]] = None

property end

Return type

float

property frame_shift

Return type

float

load(start=None, duration=None)

Return type

ndarray

with_path_prefix(path)

Return type

Features

static from_dict(data)

Return type

Features

__init__(type, num_frames, num_features, sampling_rate, start, duration, storage_type, storage_path, storage_key, recording_id=None, channels=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.features.base.FeatureSet(*args, **kwds)

Represents a feature manifest, and allows to read features for given recordings within particular channels and time ranges. It also keeps information about the feature extractor parameters used to obtain this set. When a given recording/time-range/channel is unavailable, raises a KeyError.

features: List[Features]

static from_features(features)

Return type

FeatureSet

static from_dicts(data)

Return type

FeatureSet

to_dicts()

Return type

List[dict]

with_path_prefix(path)

Return type

FeatureSet

split(num_splits, randomize=False)

Split the FeatureSet into num_splits pieces of equal size.

Parameters

• num_splits (int) – Requested number of splits.

• randomize (bool) – Optionally randomize the features order first.

Return type

List[FeatureSet]

Returns

A list of FeatureSet pieces.

find(recording_id, channel_id=0, start=0.0, duration=None, leeway=0.05)

Find and return a Features object that best satisfies the search criteria. Raise a KeyError when no such object is available.

Parameters

• recording_id (str) – str, requested recording ID.

• channel_id (int) – int, requested channel.

• start (float) – float, requested start time in seconds for the feature chunk.

• duration (Optional[float]) – optional float, requested duration in seconds for the feature chunk. By default, return everything from the start.

• leeway (float) – float, controls how strictly we have to match the requested start and duration criteria. It is necessary to keep a small positive value here (default 0.05s), as there might be differneces between the duration of recording/supervision segment, and the duration of features. The latter one is constrained to be a multiple of frame_shift, while the former can be arbitrary.

Return type

Features

Returns

a Features object satisfying the search criteria.

load(recording_id, channel_id=0, start=0.0, duration=None)

Find a Features object that best satisfies the search criteria and load the features as a numpy ndarray. Raise a KeyError when no such object is available.

Return type

ndarray

compute_global_stats(storage_path=None)

Compute the global means and standard deviations for each feature bin in the manifest. It follows the implementation in scikit-learn: https://github.com/scikit-learn/scikit-learn/blob/0fb307bf39bbdacd6ed713c00724f8f871d60370/sklearn/utils/extmath.py#L715 which follows the paper: “Algorithms for computing the sample variance: analysis and recommendations”, by Chan, Golub, and LeVeque.

Parameters

storage_path (Union[Path, str, None]) – an optional path to a file where the stats will be stored with pickle.

Return a dict of ``{‘norm_means’``{‘norm_means’

np.ndarray, ‘norm_stds’: np.ndarray}`` with the shape of the arrays equal to the number of feature bins in this manifest.

Return type

Dict[str, ndarray]

__init__(features=<factory>, _features_by_recording_id=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.features.base.FeatureSetBuilder(feature_extractor, storage, augment_fn=None)

An extended constructor for the FeatureSet. Think of it as a class wrapper for a feature extraction script. It consumes an iterable of Recordings, extracts the features specified by the FeatureExtractor config, and saves stores them on the disk.

Eventually, we plan to extend it with the capability to extract only the features in specified regions of recordings and to perform some time-domain data augmentation.

__init__(feature_extractor, storage, augment_fn=None)

Initialize self. See help(type(self)) for accurate signature.

process_and_store_recordings(recordings, output_manifest=None, num_jobs=1)

Return type

FeatureSet

lhotse.features.base.store_feature_array(feats, storage)

Store feats array on disk, using lilcom compression by default.

Parameters

• feats (ndarray) – a numpy ndarray containing features.

• storage (FeaturesWriter) – a FeaturesWriter object to use for array storage.

Return type

str

Returns

a path to the file containing the stored array.

lhotse.features.base.compute_global_stats(feature_manifests, storage_path=None)

Compute the global means and standard deviations for each feature bin in the manifest. It performs only a single pass over the data and iteratively updates the estimate of the means and variances.

We follow the implementation in scikit-learn: https://github.com/scikit-learn/scikit-learn/blob/0fb307bf39bbdacd6ed713c00724f8f871d60370/sklearn/utils/extmath.py#L715 which follows the paper: “Algorithms for computing the sample variance: analysis and recommendations”, by Chan, Golub, and LeVeque.

Parameters

• feature_manifests (Iterable[Features]) – an iterable of Features objects.

• storage_path (Union[Path, str, None]) – an optional path to a file where the stats will be stored with pickle.

Return a dict of ``{‘norm_means’``{‘norm_means’

np.ndarray, ‘norm_stds’: np.ndarray}`` with the shape of the arrays equal to the number of feature bins in this manifest.

Return type

Dict[str, ndarray]

Torchaudio feature extractors

class lhotse.features.fbank.FbankConfig(dither: float = 0.0, window_type: str = 'povey', frame_length: float = 0.025, frame_shift: float = 0.01, remove_dc_offset: bool = True, round_to_power_of_two: bool = True, energy_floor: float = 1e-10, min_duration: float = 0.0, preemphasis_coefficient: float = 0.97, raw_energy: bool = True, low_freq: float = 20.0, high_freq: float = - 400.0, num_mel_bins: int = 40, use_energy: bool = False, vtln_low: float = 100.0, vtln_high: float = - 500.0, vtln_warp: float = 1.0)

dither: float = 0.0

window_type: str = 'povey'

frame_length: float = 0.025

frame_shift: float = 0.01

remove_dc_offset: bool = True

round_to_power_of_two: bool = True

energy_floor: float = 1e-10

min_duration: float = 0.0

preemphasis_coefficient: float = 0.97

raw_energy: bool = True

low_freq: float = 20.0

high_freq: float = -400.0

num_mel_bins: int = 40

use_energy: bool = False

vtln_low: float = 100.0

vtln_high: float = -500.0

vtln_warp: float = 1.0

__init__(dither=0.0, window_type='povey', frame_length=0.025, frame_shift=0.01, remove_dc_offset=True, round_to_power_of_two=True, energy_floor=1e-10, min_duration=0.0, preemphasis_coefficient=0.97, raw_energy=True, low_freq=20.0, high_freq=- 400.0, num_mel_bins=40, use_energy=False, vtln_low=100.0, vtln_high=- 500.0, vtln_warp=1.0)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.features.fbank.Fbank(config=None)

Log Mel energy filter bank feature extractor based on torchaudio.compliance.kaldi.fbank function.

name = 'fbank'

config_type

alias of FbankConfig

feature_dim(sampling_rate)

Return type

int

static mix(features_a, features_b, energy_scaling_factor_b)

Perform feature-domain mix of two singals, a and b, and return the mixed signal.

Parameters

• features_a (ndarray) – Left-hand side (reference) signal.

• features_b (ndarray) – Right-hand side (mixed-in) signal.

• energy_scaling_factor_b (float) – A scaling factor for features_b energy. It is used to achieve a specific SNR. E.g. to mix with an SNR of 10dB when both features_a and features_b energies are 100, the features_b signal energy needs to be scaled by 0.1. Since different features (e.g. spectrogram, fbank, MFCC) require different combination of transformations (e.g. exp, log, sqrt, pow) to allow mixing of two signals, the exact place where to apply energy_scaling_factor_b to the signal is determined by the implementer.

Return type

ndarray

Returns

A mixed feature matrix.

static compute_energy(features)

Compute the total energy of a feature matrix. How the energy is computed depends on a particular type of features. It is expected that when implemented, compute_energy will never return zero.

Parameters

features (ndarray) – A feature matrix.

Return type

float

Returns

A positive float value of the signal energy.

class lhotse.features.mfcc.MfccConfig(dither: float = 0.0, window_type: str = 'povey', frame_length: float = 0.025, frame_shift: float = 0.01, remove_dc_offset: bool = True, round_to_power_of_two: bool = True, energy_floor: float = 1e-10, min_duration: float = 0.0, preemphasis_coefficient: float = 0.97, raw_energy: bool = True, low_freq: float = 20.0, high_freq: float = 0.0, num_mel_bins: int = 23, use_energy: bool = False, vtln_low: float = 100.0, vtln_high: float = - 500.0, vtln_warp: float = 1.0, cepstral_lifter: float = 22.0, num_ceps: int = 13)

dither: float = 0.0

window_type: str = 'povey'

frame_length: float = 0.025

frame_shift: float = 0.01

remove_dc_offset: bool = True

round_to_power_of_two: bool = True

energy_floor: float = 1e-10

min_duration: float = 0.0

preemphasis_coefficient: float = 0.97

raw_energy: bool = True

low_freq: float = 20.0

high_freq: float = 0.0

num_mel_bins: int = 23

use_energy: bool = False

vtln_low: float = 100.0

vtln_high: float = -500.0

vtln_warp: float = 1.0

cepstral_lifter: float = 22.0

num_ceps: int = 13

__init__(dither=0.0, window_type='povey', frame_length=0.025, frame_shift=0.01, remove_dc_offset=True, round_to_power_of_two=True, energy_floor=1e-10, min_duration=0.0, preemphasis_coefficient=0.97, raw_energy=True, low_freq=20.0, high_freq=0.0, num_mel_bins=23, use_energy=False, vtln_low=100.0, vtln_high=- 500.0, vtln_warp=1.0, cepstral_lifter=22.0, num_ceps=13)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.features.mfcc.Mfcc(config=None)

MFCC feature extractor based on torchaudio.compliance.kaldi.mfcc function.

name = 'mfcc'

config_type

alias of MfccConfig

feature_dim(sampling_rate)

Return type

int

class lhotse.features.spectrogram.SpectrogramConfig(dither: float = 0.0, window_type: str = 'povey', frame_length: float = 0.025, frame_shift: float = 0.01, remove_dc_offset: bool = True, round_to_power_of_two: bool = True, energy_floor: float = 1e-10, min_duration: float = 0.0, preemphasis_coefficient: float = 0.97, raw_energy: bool = True)

dither: float = 0.0

window_type: str = 'povey'

frame_length: float = 0.025

frame_shift: float = 0.01

remove_dc_offset: bool = True

round_to_power_of_two: bool = True

energy_floor: float = 1e-10

min_duration: float = 0.0

preemphasis_coefficient: float = 0.97

raw_energy: bool = True

__init__(dither=0.0, window_type='povey', frame_length=0.025, frame_shift=0.01, remove_dc_offset=True, round_to_power_of_two=True, energy_floor=1e-10, min_duration=0.0, preemphasis_coefficient=0.97, raw_energy=True)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.features.spectrogram.Spectrogram(config=None)

Log spectrogram feature extractor based on torchaudio.compliance.kaldi.spectrogram function.

name = 'spectrogram'

config_type

alias of SpectrogramConfig

feature_dim(sampling_rate)

Return type

int

static mix(features_a, features_b, energy_scaling_factor_b)

Perform feature-domain mix of two singals, a and b, and return the mixed signal.

Parameters

• features_a (ndarray) – Left-hand side (reference) signal.

• features_b (ndarray) – Right-hand side (mixed-in) signal.

• energy_scaling_factor_b (float) – A scaling factor for features_b energy. It is used to achieve a specific SNR. E.g. to mix with an SNR of 10dB when both features_a and features_b energies are 100, the features_b signal energy needs to be scaled by 0.1. Since different features (e.g. spectrogram, fbank, MFCC) require different combination of transformations (e.g. exp, log, sqrt, pow) to allow mixing of two signals, the exact place where to apply energy_scaling_factor_b to the signal is determined by the implementer.

Return type

ndarray

Returns

A mixed feature matrix.

static compute_energy(features)

Compute the total energy of a feature matrix. How the energy is computed depends on a particular type of features. It is expected that when implemented, compute_energy will never return zero.

Parameters

features (ndarray) – A feature matrix.

Return type

float

Returns

A positive float value of the signal energy.

Feature storage

class lhotse.features.io.FeaturesWriter

FeaturesWriter defines the interface of how to store numpy arrays in a particular storage backend. This backend could either be:

• separate files on a local filesystem;

• a single file with multiple arrays;

• cloud storage;

• etc.

Each class inheriting from FeaturesWriter must define:

• the write() method, which defines the storing operation

  (accepts a key used to place the value array in the storage);

• the storage_path() property, which is either a common directory for the files,

  the name of the file storing multiple arrays, name of the cloud bucket, etc.

• the name() property that is unique to this particular storage mechanism -

  it is stored in the features manifests (metadata) and used to automatically deduce the backend when loading the features.

Each FeaturesWriter can also be used as a context manager, as some implementations might need to free a resource after the writing is finalized. By default nothing happens in the context manager functions, and this can be modified by the inheriting subclasses.

Example:

with MyWriter(‘some/path’) as storage:

extractor.extract_from_recording_and_store(recording, storage)

The features loading must be defined separately in a class inheriting from FeaturesReader.

abstract property name

Return type

str

abstract property storage_path

Return type

str

abstract write(key, value)

Return type

str

class lhotse.features.io.FeaturesReader

FeaturesReader defines the interface of how to load numpy arrays from a particular storage backend. This backend could either be:

• separate files on a local filesystem;

• a single file with multiple arrays;

• cloud storage;

• etc.

Each class inheriting from FeaturesReader must define:

• the read() method, which defines the loading operation

  (accepts the key to locate the array in the storage and return it). The read method should support selecting only a subset of the feature matrix, with the bounds expressed as arguments left_offset_frames and right_offset_frames. It’s up to the Reader implementation to load only the required part or trim it to that range only after loading. It is assumed that the time dimension is always the first one.

• the name() property that is unique to this particular storage mechanism -

  it is stored in the features manifests (metadata) and used to automatically deduce the backend when loading the features.

The features writing must be defined separately in a class inheriting from FeaturesWriter.

abstract property name

Return type

str

abstract read(key, left_offset_frames=0, right_offset_frames=None)

Return type

ndarray

lhotse.features.io.available_storage_backends()

Return type

List[str]

lhotse.features.io.register_reader(cls)

Decorator used to add a new FeaturesReader to Lhotse’s registry.

Example:

@register_reader class MyFeatureReader(FeatureReader):

…

lhotse.features.io.register_writer(cls)

Decorator used to add a new FeaturesWriter to Lhotse’s registry.

Example:

@register_writer class MyFeatureWriter(FeatureWriter):

…

lhotse.features.io.get_reader(name)

Find a FeaturesReader sub-class that corresponds to the provided name and return its type.

Example:

reader_type = get_reader(“lilcom_files”) reader = reader_type(“/storage/features/”)

Return type

Type[FeaturesReader]

lhotse.features.io.get_writer(name)

Find a FeaturesWriter sub-class that corresponds to the provided name and return its type.

Example:

writer_type = get_writer(“lilcom_files”) writer = writer_type(“/storage/features/”)

Return type

Type[FeaturesWriter]

class lhotse.features.io.LilcomFilesReader(storage_path, *args, **kwargs)

Reads Lilcom-compressed files from a directory on the local filesystem. storage_path corresponds to the directory path; storage_key for each utterance is the name of the file in that directory.

name = 'lilcom_files'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

read(key, left_offset_frames=0, right_offset_frames=None)

Return type

ndarray

class lhotse.features.io.LilcomFilesWriter(storage_path, tick_power=- 5, *args, **kwargs)

Writes Lilcom-compressed files to a directory on the local filesystem. storage_path corresponds to the directory path; storage_key for each utterance is the name of the file in that directory.

name = 'lilcom_files'

__init__(storage_path, tick_power=- 5, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

property storage_path

Return type

str

write(key, value)

Return type

str

class lhotse.features.io.NumpyFilesReader(storage_path, *args, **kwargs)

Reads non-compressed numpy arrays from files in a directory on the local filesystem. storage_path corresponds to the directory path; storage_key for each utterance is the name of the file in that directory.

name = 'numpy_files'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

read(key, left_offset_frames=0, right_offset_frames=None)

Return type

ndarray

class lhotse.features.io.NumpyFilesWriter(storage_path, *args, **kwargs)

Writes non-compressed numpy arrays to files in a directory on the local filesystem. storage_path corresponds to the directory path; storage_key for each utterance is the name of the file in that directory.

name = 'numpy_files'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

property storage_path

Return type

str

write(key, value)

Return type

str

lhotse.features.io.lookup_cache_or_open(storage_path)

Helper internal function used in HDF5 readers. It opens the HDF files and keeps their handles open in a global program cache to avoid excessive amount of syscalls when the *Reader class is instantiated and destroyed in a loop repeatedly (frequent use-case).

The file handles can be freed at any time by calling close_cached_file_handles().

lhotse.features.io.close_cached_file_handles()

Closes the cached file handles in lookup_cache_or_open (see its docs for more details).

Return type

None

class lhotse.features.io.NumpyHdf5Reader(storage_path, *args, **kwargs)

Reads non-compressed numpy arrays from a HDF5 file with a “flat” layout. Each array is stored as a separate HDF Dataset because their shapes (numbers of frames) may vary. storage_path corresponds to the HDF5 file path; storage_key for each utterance is the key corresponding to the array (i.e. HDF5 “Group” name).

name = 'numpy_hdf5'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

read(key, left_offset_frames=0, right_offset_frames=None)

Return type

ndarray

class lhotse.features.io.NumpyHdf5Writer(storage_path, *args, **kwargs)

Writes non-compressed numpy arrays to a HDF5 file with a “flat” layout. Each array is stored as a separate HDF Dataset because their shapes (numbers of frames) may vary. storage_path corresponds to the HDF5 file path; storage_key for each utterance is the key corresponding to the array (i.e. HDF5 “Group” name).

name = 'numpy_hdf5'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

property storage_path

Return type

str

write(key, value)

Return type

str

close()

Return type

None

class lhotse.features.io.LilcomHdf5Reader(storage_path, *args, **kwargs)

Reads lilcom-compressed numpy arrays from a HDF5 file with a “flat” layout. Each array is stored as a separate HDF Dataset because their shapes (numbers of frames) may vary. storage_path corresponds to the HDF5 file path; storage_key for each utterance is the key corresponding to the array (i.e. HDF5 “Group” name).

name = 'lilcom_hdf5'

__init__(storage_path, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

read(key, left_offset_frames=0, right_offset_frames=None)

Return type

ndarray

class lhotse.features.io.LilcomHdf5Writer(storage_path, tick_power=- 5, *args, **kwargs)

Writes lilcom-compressed numpy arrays to a HDF5 file with a “flat” layout. Each array is stored as a separate HDF Dataset because their shapes (numbers of frames) may vary. storage_path corresponds to the HDF5 file path; storage_key for each utterance is the key corresponding to the array (i.e. HDF5 “Group” name).

name = 'lilcom_hdf5'

__init__(storage_path, tick_power=- 5, *args, **kwargs)

Initialize self. See help(type(self)) for accurate signature.

property storage_path

Return type

str

write(key, value)

Return type

str

close()

Return type

None

Feature-domain mixing

class lhotse.features.mixer.FeatureMixer(feature_extractor, base_feats, frame_shift, padding_value=- 1000.0)

Utility class to mix multiple feature matrices into a single one. It should be instantiated separately for each mixing session (i.e. each MixedCut will create a separate FeatureMixer to mix its tracks). It is initialized with a numpy array of features (typically float32) that represents the “reference” signal for the mix. Other signals can be mixed to it with different time offsets and SNRs using the add_to_mix method. The time offset is relative to the start of the reference signal (only positive values are supported). The SNR is relative to the energy of the signal used to initialize the FeatureMixer.

It relies on the FeatureExtractor to have defined mix and compute_energy methods, so that the FeatureMixer knows how to scale and add two feature matrices together.

__init__(feature_extractor, base_feats, frame_shift, padding_value=- 1000.0)

Parameters

• feature_extractor (FeatureExtractor) – The FeatureExtractor instance that specifies how to mix the features.

• base_feats (ndarray) – The features used to initialize the FeatureMixer are a point of reference in terms of energy and offset for all features mixed into them.

• frame_shift (float) – Required to correctly compute offset and padding during the mix.

• padding_value (float) – The value used to pad the shorter features during the mix. This value is adequate only for log space features. For non-log space features, e.g. energies, use either 0 or a small positive value like 1e-5.

property num_features

property unmixed_feats

Return a numpy ndarray with the shape (num_tracks, num_frames, num_features), where each track’s feature matrix is padded and scaled adequately to the offsets and SNR used in add_to_mix call.

Return type

ndarray

property mixed_feats

Return a numpy ndarray with the shape (num_frames, num_features) - a mono mixed feature matrix of the tracks supplied with add_to_mix calls.

Return type

ndarray

add_to_mix(feats, snr=None, offset=0.0)

Add feature matrix of a new track into the mix. :type feats: ndarray :param feats: A 2D feature matrix to be mixed in. :type snr: Optional[float] :param snr: Signal-to-noise ratio, assuming feats represents noise (positive SNR - lower feats energy, negative SNR - higher feats energy) :type offset: float :param offset: How many seconds to shift feats in time. For mixing, the signal will be padded before the start with low energy values.

Augmentation

Cuts

Data structures and tools used to create training/testing examples.

class lhotse.cut.CutUtilsMixin

A mixin class for cuts which contains all the methods that share common implementations.

Note: Ideally, this would’ve been an abstract base class specifying the common interface, but ABC’s do not mix well with dataclasses in Python. It is possible we’ll ditch the dataclass for cuts in the future and make this an ABC instead.

property trimmed_supervisions

Return the supervisions in this Cut that have modified time boundaries so as not to exceed the Cut’s start or end.

Note that when cut.supervisions is called, the supervisions may have negative start values that indicate the supervision actually begins before the cut, or end values that exceed the Cut’s duration (it means the supervision continued in the original recording after the Cut’s ending).

Return type

List[SupervisionSegment]

mix(other, offset_other_by=0.0, snr=None)

Refer to mix() documentation.

Return type

MixedCut

append(other, snr=None)

Append the other Cut after the current Cut. Conceptually the same as mix but with an offset matching the current cuts length. Optionally scale down (positive SNR) or scale up (negative SNR) the other cut. Returns a MixedCut, which only keeps the information about the mix; actual mixing is performed during the call to load_features.

Return type

MixedCut

compute_features(extractor, augment_fn=None)

Compute the features from this cut. This cut has to be able to load audio.

Parameters

• extractor (FeatureExtractor) – a FeatureExtractor instance used to compute the features.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – optional WavAugmenter instance for audio augmentation.

Return type

ndarray

Returns

a numpy ndarray with the computed features.

plot_audio()

Display a plot of the waveform. Requires matplotlib to be installed.

play_audio()

Display a Jupyter widget that allows to listen to the waveform. Works only in Jupyter notebook/lab or similar (e.g. Colab).

plot_features()

Display the feature matrix as an image. Requires matplotlib to be installed.

speakers_feature_mask(min_speaker_dim=None, speaker_to_idx_map=None)

Return a matrix of per-speaker activity in a cut. The matrix shape is (num_speakers, num_frames), and its values are 0 for nonspeech frames and 1 for speech frames for each respective speaker.

This is somewhat inspired by the TS-VAD setup: https://arxiv.org/abs/2005.07272

Parameters

• min_speaker_dim (Optional[int]) – optional int, when specified it will enforce that the matrix shape is at least that value (useful for datasets like CHiME 6 where the number of speakers is always 4, but some cuts might have less speakers than that).

• speaker_to_idx_map (Optional[Dict[str, int]]) – optional dict mapping speaker names (strings) to their global indices (ints). Useful when you want to preserve the order of the speakers (e.g. speaker XYZ is always mapped to index 2)

Return type

ndarray

speakers_audio_mask(min_speaker_dim=None, speaker_to_idx_map=None)

Return a matrix of per-speaker activity in a cut. The matrix shape is (num_speakers, num_samples), and its values are 0 for nonspeech samples and 1 for speech samples for each respective speaker.

This is somewhat inspired by the TS-VAD setup: https://arxiv.org/abs/2005.07272

Parameters

• min_speaker_dim (Optional[int]) – optional int, when specified it will enforce that the matrix shape is at least that value (useful for datasets like CHiME 6 where the number of speakers is always 4, but some cuts might have less speakers than that).

• speaker_to_idx_map (Optional[Dict[str, int]]) – optional dict mapping speaker names (strings) to their global indices (ints). Useful when you want to preserve the order of the speakers (e.g. speaker XYZ is always mapped to index 2)

Return type

ndarray

supervisions_feature_mask()

Return a 1D numpy array with value 1 for frames covered by at least one supervision, and 0 for frames not covered by any supervision.

Return type

ndarray

supervisions_audio_mask()

Return a 1D numpy array with value 1 for samples covered by at least one supervision, and 0 for samples not covered by any supervision.

Return type

ndarray

with_id(id_)

Return a copy of the Cut with a new ID.

Return type

Union[Cut, MixedCut, PaddingCut]

class lhotse.cut.Cut(id: str, start: float, duration: float, channel: int, supervisions: List[lhotse.supervision.SupervisionSegment] = <factory>, features: Optional[lhotse.features.base.Features] = None, recording: Optional[lhotse.audio.Recording] = None)

A Cut is a single “segment” that we’ll train on. It contains the features corresponding to a piece of a recording, with zero or more SupervisionSegments.

The SupervisionSegments indicate which time spans of the Cut contain some kind of supervision information: e.g. transcript, speaker, language, etc. The regions without a corresponding SupervisionSegment may contain anything - usually we assume it’s either silence or some kind of noise.

Note: The SupervisionSegment time boundaries are relative to the beginning of the cut. E.g. if the underlying Recording starts at 0s (always true), the Cut starts at 100s, and the SupervisionSegment starts at 3s, it means that in the Recording the supervision actually started at 103s. In some cases, the supervision might have a negative start, or a duration exceeding the duration of the Cut; this means that the supervision in the recording extends beyond the Cut.

id: str

start: Seconds

duration: Seconds

channel: int

supervisions: List[SupervisionSegment]

features: Optional[lhotse.features.base.Features] = None

recording: Optional[lhotse.audio.Recording] = None

property recording_id

Return type

str

property end

Return type

float

property has_features

Return type

bool

property has_recording

Return type

bool

property frame_shift

Return type

Optional[float]

property num_frames

Return type

Optional[int]

property num_samples

Return type

Optional[int]

property num_features

Return type

Optional[int]

property features_type

Return type

Optional[str]

property sampling_rate

Return type

int

load_features()

Load the features from the underlying storage and cut them to the relevant [begin, duration] region of the current Cut.

Return type

Optional[ndarray]

load_audio()

Load the audio by locating the appropriate recording in the supplied RecordingSet. The audio is trimmed to the [begin, end] range specified by the Cut.

Return type

Optional[ndarray]

Returns

a numpy ndarray with audio samples, with shape (1 <channel>, N <samples>)

compute_and_store_features(extractor, storage, augment_fn=None, *args, **kwargs)

Compute the features from this cut, store them on disk, and attach a feature manifest to this cut. This cut has to be able to load audio.

Parameters

• extractor (FeatureExtractor) – a FeatureExtractor instance used to compute the features.

• output_dir – the directory where the computed features will be stored.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – an optional callable used for audio augmentation.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a new Cut instance with a Features manifest attached to it.

truncate(*, offset=0.0, duration=None, keep_excessive_supervisions=True, preserve_id=False, _supervisions_index=None)

Returns a new Cut that is a sub-region of the current Cut.

Note that no operation is done on the actual features - it’s only during the call to load_features() when the actual changes happen (a subset of features is loaded).

Parameters

• offset (float) – float (seconds), controls the start of the new cut relative to the current Cut’s start. E.g., if the current Cut starts at 10.0, and offset is 2.0, the new start is 12.0.

• duration (Optional[float]) – optional float (seconds), controls the duration of the resulting Cut. By default, the duration is (end of the cut before truncation) - (offset).

• keep_excessive_supervisions (bool) – bool. Since trimming may happen inside a SupervisionSegment, the caller has an option to either keep or discard such supervisions.

• preserve_id (bool) – bool. Should the truncated cut keep the same ID or get a new, random one.

• _supervisions_index (Optional[Dict[str, IntervalTree]]) – when passed, allows to speed up processing of Cuts with a very large number of supervisions. Intended as an internal parameter.

Return type

Cut

Returns

a new Cut instance. If the current Cut is shorter than the duration, return None.

pad(duration)

Return a new MixedCut, padded to duration seconds with zeros in the recording, and low-energy values in each feature bin.

Parameters

duration (float) – The cut’s minimal duration after padding.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a padded MixedCut if duration is greater than this cut’s duration, otherwise self.

map_supervisions(transform_fn)

Modify the SupervisionSegments by transform_fn of this Cut.

Parameters

transform_fn (Callable[[SupervisionSegment], SupervisionSegment]) – a function that modifies a supervision as an argument.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a modified Cut.

filter_supervisions(predicate)

Modify cut to store only supervisions accepted by predicate

Example:

>>> cut = cut.filter_supervisions(lambda s: s.id in supervision_ids)
>>> cut = cut.filter_supervisions(lambda s: s.duration < 5.0)
>>> cut = cut.filter_supervisions(lambda s: s.text is not None)

Parameters

predicate (Callable[[SupervisionSegment], bool]) – A callable that accepts SupervisionSegment and returns bool

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a modified Cut

static from_dict(data)

Return type

Cut

with_features_path_prefix(path)

Return type

Cut

with_recording_path_prefix(path)

Return type

Cut

__init__(id, start, duration, channel, supervisions=<factory>, features=None, recording=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.cut.PaddingCut(id: str, duration: float, sampling_rate: int, use_log_energy: bool, num_frames: Optional[int] = None, num_features: Optional[int] = None, num_samples: Optional[int] = None)

This represents a cut filled with zeroes in the time domain, or low energy/log-energy values in the frequency domain. It’s used to make training samples evenly sized (same duration/number of frames).

id: str

duration: Seconds

sampling_rate: int

use_log_energy: bool

num_frames: Optional[int] = None

num_features: Optional[int] = None

num_samples: Optional[int] = None

property start

Return type

float

property end

Return type

float

property supervisions

property has_features

Return type

bool

property has_recording

Return type

bool

property frame_shift

load_features(*args, **kwargs)

Return type

Optional[ndarray]

load_audio(*args, **kwargs)

Return type

Optional[ndarray]

truncate(*, offset=0.0, duration=None, keep_excessive_supervisions=True, preserve_id=False, **kwargs)

Return type

PaddingCut

pad(duration)

Create a new PaddingCut with duration when its longer than this Cuts duration. Helper function used in batch cut padding.

Parameters

duration (float) – The cuts minimal duration after padding.

Return type

PaddingCut

Returns

self or a new PaddingCut, depending on duration.

compute_and_store_features(extractor, *args, **kwargs)

Returns a new PaddingCut with updates information about the feature dimension and number of feature frames, depending on the extractor properties.

Return type

Union[Cut, MixedCut, PaddingCut]

map_supervisions(transform_fn)

Just for consistency with Cut and MixedCut.

Parameters

transform_fn (Callable[[Any], Any]) – a dummy function that would be never called actually.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

the PaddingCut itself.

filter_supervisions(predicate)

Just for consistency with Cut and MixedCut.

Parameters

predicate (Callable[[SupervisionSegment], bool]) – A callable that accepts SupervisionSegment and returns bool

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a modified Cut

static from_dict(data)

Return type

PaddingCut

with_features_path_prefix(path)

Return type

PaddingCut

with_recording_path_prefix(path)

Return type

PaddingCut

__init__(id, duration, sampling_rate, use_log_energy, num_frames=None, num_features=None, num_samples=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.cut.MixTrack(cut: Union[lhotse.cut.Cut, lhotse.cut.PaddingCut], offset: float = 0.0, snr: Optional[float] = None)

Represents a single track in a mix of Cuts. Points to a specific Cut and holds information on how to mix it with other Cuts, relative to the first track in a mix.

cut: Union[Cut, PaddingCut]

offset: float = 0.0

snr: Optional[float] = None

static from_dict(data)

__init__(cut, offset=0.0, snr=None)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.cut.MixedCut(id: str, tracks: List[lhotse.cut.MixTrack])

Represents a Cut that’s created from other Cuts via mix or append operations. The actual mixing operations are performed upon loading the features into memory. In order to load the features, it needs to access the CutSet object that holds the “ingredient” cuts, as it only holds their IDs (“pointers”). The SNR and offset of all the tracks are specified relative to the first track.

id: str

tracks: List[MixTrack]

property supervisions

Lists the supervisions of the underlying source cuts. Each segment start time will be adjusted by the track offset.

Return type

List[SupervisionSegment]

property start

Return type

float

property end

Return type

float

property duration

Return type

float

property has_features

Return type

bool

property has_recording

Return type

bool

property num_frames

Return type

Optional[int]

property frame_shift

Return type

Optional[float]

property sampling_rate

Return type

Optional[int]

property num_samples

Return type

Optional[int]

property num_features

Return type

Optional[int]

property features_type

Return type

Optional[str]

truncate(*, offset=0.0, duration=None, keep_excessive_supervisions=True, preserve_id=False, _supervisions_index=None)

Returns a new MixedCut that is a sub-region of the current MixedCut. This method truncates the underlying Cuts and modifies their offsets in the mix, as needed. Tracks that do not fit in the truncated cut are removed.

Note that no operation is done on the actual features - it’s only during the call to load_features() when the actual changes happen (a subset of features is loaded).

Parameters

• offset (float) – float (seconds), controls the start of the new cut relative to the current MixedCut’s start.

• duration (Optional[float]) – optional float (seconds), controls the duration of the resulting MixedCut. By default, the duration is (end of the cut before truncation) - (offset).

• keep_excessive_supervisions (bool) – bool. Since trimming may happen inside a SupervisionSegment, the caller has an option to either keep or discard such supervisions.

• preserve_id (bool) – bool. Should the truncated cut keep the same ID or get a new, random one.

Return type

MixedCut

Returns

a new MixedCut instance.

pad(duration)

Return a new MixedCut, padded to duration seconds with zeros in the recording, and low-energy values in each feature bin.

Parameters

duration (float) – The cut’s minimal duration after padding.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a padded MixedCut if duration is greater than this cut’s duration, otherwise self.

load_features(mixed=True)

Loads the features of the source cuts and mixes them on-the-fly.

Parameters

mixed (bool) – when True (default), returns a 2D array of features mixed in the feature domain. Otherwise returns a 3D array with the first dimension equal to the number of tracks.

Return type

Optional[ndarray]

Returns

A numpy ndarray with features and with shape (num_frames, num_features), or (num_tracks, num_frames, num_features)

load_audio(mixed=True)

Loads the audios of the source cuts and mix them on-the-fly.

Parameters

mixed (bool) – When True (default), returns a mono mix of the underlying tracks. Otherwise returns a numpy array with the number of channels equal to the number of tracks.

Return type

Optional[ndarray]

Returns

A numpy ndarray with audio samples and with shape (num_channels, num_samples)

plot_tracks_features()

Display the feature matrix as an image. Requires matplotlib to be installed.

plot_tracks_audio()

Display plots of the individual tracks’ waveforms. Requires matplotlib to be installed.

compute_and_store_features(extractor, storage, augment_fn=None, mix_eagerly=True)

Compute the features from this cut, store them on disk, and create a new Cut object with the feature manifest attached. This cut has to be able to load audio.

Parameters

• extractor (FeatureExtractor) – a FeatureExtractor instance used to compute the features.

• storage (FeaturesWriter) – a FeaturesWriter instance used to store the features.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – an optional callable used for audio augmentation.

• mix_eagerly (bool) – when False, extract and store the features for each track separately, and mix them dynamically when loading the features. When True, mix the audio first and store the mixed features, returning a new Cut instance with the same ID. The returned Cut will not have a Recording attached.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a new Cut instance if mix_eagerly is True, or returns self with each of the tracks containing the Features manifests.

map_supervisions(transform_fn)

Modify the SupervisionSegments by transform_fn of this MixedCut.

Parameters

transform_fn (Callable[[SupervisionSegment], SupervisionSegment]) – a function that modifies a supervision as an argument.

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a modified MixedCut.

filter_supervisions(predicate)

Modify cut to store only supervisions accepted by predicate

Example:

>>> cut = cut.filter_supervisions(lambda s: s.id in supervision_ids)
>>> cut = cut.filter_supervisions(lambda s: s.duration < 5.0)
>>> cut = cut.filter_supervisions(lambda s: s.text is not None)

Parameters

predicate (Callable[[SupervisionSegment], bool]) – A callable that accepts SupervisionSegment and returns bool

Return type

Union[Cut, MixedCut, PaddingCut]

Returns

a modified Cut

static from_dict(data)

Return type

MixedCut

with_features_path_prefix(path)

Return type

MixedCut

with_recording_path_prefix(path)

Return type

MixedCut

__init__(id, tracks)

Initialize self. See help(type(self)) for accurate signature.

class lhotse.cut.CutSet(*args, **kwds)

CutSet combines features with their corresponding supervisions. It may have wider span than the actual supervisions, provided the features for the whole span exist. It is the basic building block of PyTorch-style Datasets for speech/audio processing tasks.

cuts: Dict[str, AnyCut]

property mixed_cuts

Return type

Dict[str, MixedCut]

property simple_cuts

Return type

Dict[str, Cut]

property ids

Return type

Iterable[str]

property speakers

Return type

FrozenSet[str]

static from_cuts(cuts)

Return type

CutSet

static from_manifests(recordings=None, supervisions=None, features=None)

Create a CutSet from any combination of supervision, feature and recording manifests. At least one of recording_set or feature_set is required. The Cut boundaries correspond to those found in the feature_set, when available, otherwise to those found in the recording_set When a supervision_set is provided, we’ll attach to the Cut all supervisions that have a matching recording ID and are fully contained in the Cut’s boundaries.

Return type

CutSet

static from_dicts(data)

Return type

CutSet

to_dicts()

Return type

List[dict]

describe()

Print a message describing details about the CutSet - the number of cuts and the duration statistics, including the total duration and the percentage of speech segments.

Example output:

Cuts count: 547 Total duration (hours): 326.4 Speech duration (hours): 79.6 (24.4%) *** Duration statistics (seconds): mean 2148.0 std 870.9 min 477.0 25% 1523.0 50% 2157.0 75% 2423.0 max 5415.0 dtype: float64

Return type

None

split(num_splits, randomize=False)

Split the CutSet into num_splits pieces of equal size.

Parameters

• num_splits (int) – Requested number of splits.

• randomize (bool) – Optionally randomize the cuts order first.

Return type

List[CutSet]

Returns

A list of CutSet pieces.

subset(supervision_ids=None)

Return a new CutSet with Cuts containing only supervision_ids.

Cuts without supervisions are removed.

Example:

>>> cuts = CutSet.from_yaml('path/to/cuts')
>>> train_set = cuts.subset(supervision_ids=train_ids)
>>> test_set = cuts.subset(supervision_ids=test_ids)

Parameters

supervision_ids (Optional[Iterable[str]]) – List of supervision_ids to keep

Return type

CutSet

Returns

a CutSet with filtered supervisions

filter_supervisions(predicate)

Return a new CutSet with Cuts containing only SupervisionSegments satisfying predicate

Cuts without supervisions are preserved

Example:

>>> cuts = CutSet.from_yaml('path/to/cuts')
>>> at_least_five_second_supervisions = cuts.filter_supervisions(lambda s: s.duration >= 5)

Parameters

predicate (Callable[[SupervisionSegment], bool]) – A callable that accepts SupervisionSegment and returns bool

Return type

CutSet

Returns

a CutSet with filtered supervisions

filter(predicate)

Return a new CutSet with the Cuts that satisfy the predicate.

Parameters

predicate (Callable[[Union[Cut, MixedCut, PaddingCut]], bool]) – a function that takes a cut as an argument and returns bool.

Return type

CutSet

Returns

a filtered CutSet.

trim_to_supervisions()

Return a new CutSet with Cuts that have identical spans as their supervisions.

Return type

CutSet

Returns

a CutSet.

trim_to_unsupervised_segments()

Return a new CutSet with Cuts created from segments that have no supervisions (likely silence or noise).

Return type

CutSet

Returns

a CutSet.

mix_same_recording_channels()

Find cuts that come from the same recording and have matching start and end times, but represent different channels. Then, mix them together (in matching groups) and return a new CutSet that contains their mixes. This is useful for processing microphone array recordings.

It is intended to be used as the first operation after creating a new CutSet (but might also work in other circumstances, e.g. if it was cut to windows first).

Example:

>>> ami = prepare_ami('path/to/ami')
>>> cut_set = CutSet.from_manifests(recordings=ami['train']['recordings'])
>>> multi_channel_cut_set = cut_set.mix_same_recording_channels()

In the AMI example, the multi_channel_cut_set will yield MixedCuts that hold all single-channel Cuts together.

Return type

CutSet

sort_by_duration(ascending=False)

Sort the CutSet according to cuts duration. Descending by default.

Return type

CutSet

index_supervisions(index_mixed_tracks=False)

Create a two-level index of supervision segments. It is a mapping from a Cut’s ID to an interval tree that contains the supervisions of that Cut.

The interval tree can be efficiently queried for overlapping and/or enveloping segments. It helps speed up some operations on Cuts of very long recordings (1h+) that contain many supervisions.

Parameters

index_mixed_tracks (bool) – Should the tracks of MixedCut’s be indexed as additional, separate entries.

Return type

Dict[str, IntervalTree]

Returns

a mapping from Cut ID to an interval tree of SupervisionSegments.

pad(duration=None)

Return a new CutSet with Cuts padded to duration in seconds. Cuts longer than duration will not be affected. Cuts will be padded to the right (i.e. after the signal). :type duration: Optional[float] :param duration: The cuts minimal duration after padding. When not specified, we’ll choose the duration of the longest cut in the CutSet. :rtype: CutSet :return: A padded CutSet.

truncate(max_duration, offset_type, keep_excessive_supervisions=True, preserve_id=False)

Return a new CutSet with the Cuts truncated so that their durations are at most max_duration. Cuts shorter than max_duration will not be changed. :type max_duration: float :param max_duration: float, the maximum duration in seconds of a cut in the resulting manifest. :type offset_type: str :param offset_type: str, can be: - ‘start’ => cuts are truncated from their start; - ‘end’ => cuts are truncated from their end minus max_duration; - ‘random’ => cuts are truncated randomly between their start and their end minus max_duration :type keep_excessive_supervisions: bool :param keep_excessive_supervisions: bool. When a cut is truncated in the middle of a supervision segment, should the supervision be kept. :type preserve_id: bool :param preserve_id: bool. Should the truncated cut keep the same ID or get a new, random one. :rtype: CutSet :return: a new CutSet instance with truncated cuts.

cut_into_windows(duration, keep_excessive_supervisions=True)

Return a new CutSet, made by traversing each Cut in windows of duration seconds and creating new Cut out of them.

The last window might have a shorter duration if there was not enough audio, so you might want to use either .filter() or .pad() afterwards to obtain a uniform duration CutSet.

Parameters

• duration (float) – Desired duration of the new cuts in seconds.

• keep_excessive_supervisions (bool) – bool. When a cut is truncated in the middle of a supervision segment, should the supervision be kept.

Return type

CutSet

Returns

a new CutSet with cuts made from shorter duration windows.

compute_and_store_features(extractor, storage, augment_fn=None, executor=None, mix_eagerly=True)

Modify the current CutSet with by extracting features and attaching the feature manifests to the cuts.

Parameters

• extractor (FeatureExtractor) – A FeatureExtractor instance (either Lhotse’s built-in or a custom implementation).

• storage (FeaturesWriter) – A FeaturesWriter instance used to store the features.

• augment_fn (Optional[Callable[[ndarray, int], ndarray]]) – an optional callable used for audio augmentation.

• executor (Optional[Any]) – when provided, will be used to parallelize the feature extraction process. Any executor satisfying the standard concurrent.futures interface will be suitable; e.g. ProcessPoolExecutor, ThreadPoolExecutor, or dask.Client for distributed task execution (see: https://docs.dask.org/en/latest/futures.html?highlight=Client#start-dask-client)

• mix_eagerly (bool) – Related to how the features are extracted for MixedCut instances, if any are present. When False, extract and store the features for each track separately, and mix them dynamically when loading the features. When True, mix the audio first and store the mixed features, returning a new Cut instance with the same ID. The returned Cut will not have a Recording attached.

Return type

CutSet

Returns

a new CutSet instance with the same Cut``s, but with attached ``Features objects

compute_global_feature_stats(storage_path=None, max_cuts=None)

Compute the global means and standard deviations for each feature bin in the manifest. It follows the implementation in scikit-learn: https://github.com/scikit-learn/scikit-learn/blob/0fb307bf39bbdacd6ed713c00724f8f871d60370/sklearn/utils/extmath.py#L715 which follows the paper: “Algorithms for computing the sample variance: analysis and recommendations”, by Chan, Golub, and LeVeque.

Parameters

• storage_path (Union[Path, str, None]) – an optional path to a file where the stats will be stored with pickle.

• max_cuts (Optional[int]) – optionally, limit the number of cuts used for stats estimation. The cuts will be selected randomly in that case.

Return a dict of ``{‘norm_means’``{‘norm_means’

np.ndarray, ‘norm_stds’: np.ndarray}`` with the shape of the arrays equal to the number of feature bins in this manifest.

Return type

Dict[str, ndarray]

with_features_path_prefix(path)

Return type

CutSet

with_recording_path_prefix(path)

Return type

CutSet

map_supervisions(transform_fn)

Modify the SupervisionSegments by transform_fn in this CutSet.

Parameters

transform_fn (Callable[[SupervisionSegment], SupervisionSegment]) – a function that modifies a supervision as an argument.

Return type

CutSet

Returns

a new, modified CutSet.

transform_text(transform_fn)

Return a copy of this CutSet with all SupervisionSegments text transformed with transform_fn. Useful for text normalization, phonetic transcription, etc.

Parameters

transform_fn (Callable[[str], str]) – a function that accepts a string and returns a string.

Return type

CutSet

Returns

a new, modified CutSet.

__init__(cuts)

Initialize self. See help(type(self)) for accurate signature.

lhotse.cut.make_windowed_cuts_from_features(feature_set, cut_duration, cut_shift=None, keep_shorter_windows=False)

Converts a FeatureSet to a CutSet by traversing each Features object in - possibly overlapping - windows, and creating a Cut out of that area. By default, the last window in traversal will be discarded if it cannot satisfy the cut_duration requirement.

Parameters

• feature_set (FeatureSet) – a FeatureSet object.

• cut_duration (float) – float, duration of created Cuts in seconds.

• cut_shift (Optional[float]) – optional float, specifies how many seconds are in between the starts of consecutive windows. Equals cut_duration by default.

• keep_shorter_windows (bool) – bool, when True, the last window will be used to create a Cut even if its duration is shorter than cut_duration.

Return type

CutSet

Returns

a CutSet object.

lhotse.cut.mix(reference_cut, mixed_in_cut, offset=0, snr=None)

Overlay, or mix, two cuts. Optionally the mixed_in_cut may be shifted by offset seconds and scaled down (positive SNR) or scaled up (negative SNR). Returns a MixedCut, which contains both cuts and the mix information. The actual feature mixing is performed during the call to MixedCut.load_features().

Parameters

• reference_cut (Union[Cut, MixedCut, PaddingCut]) – The reference cut for the mix - offset and snr are specified w.r.t this cut.

• mixed_in_cut (Union[Cut, MixedCut, PaddingCut]) – The mixed-in cut - it will be offset and rescaled to match the offset and snr parameters.

• offset (float) – How many seconds to shift the mixed_in_cut w.r.t. the reference_cut.

• snr (Optional[float]) – Desired SNR of the right_cut w.r.t. the left_cut in the mix.

Return type

MixedCut

Returns

A MixedCut instance.

lhotse.cut.append(left_cut, right_cut, snr=None)

Helper method for functional-style appending of Cuts.

Return type

MixedCut

lhotse.cut.mix_cuts(cuts)

Return a MixedCut that consists of the input Cuts mixed with each other as-is.

Return type

MixedCut

lhotse.cut.append_cuts(cuts)

Return a MixedCut that consists of the input Cuts appended to each other as-is.

Return type

Union[Cut, MixedCut, PaddingCut]

Recipes

Convenience methods used to prepare recording and supervision manifests for standard corpora.

Kaldi conversion

Convenience methods used to interact with Kaldi data directories.

lhotse.kaldi.load_kaldi_data_dir(path, sampling_rate)

Load a Kaldi data directory and convert it to a Lhotse RecordingSet and SupervisionSet manifests. For this to work, at least the wav.scp file must exist. SupervisionSet is created only when a segments file exists. All the other files (text, utt2spk, etc.) are optional, and some of them might not be handled yet. In particular, feats.scp files are ignored.

Return type

Tuple[RecordingSet, Optional[SupervisionSet]]

lhotse.kaldi.export_to_kaldi(recordings, supervisions, output_dir)

Export a pair of RecordingSet and SupervisionSet to a Kaldi data directory. Currently, it only supports single-channel recordings that have a single AudioSource.

The RecordingSet and SupervisionSet must be compatible, i.e. it must be possible to create a CutSet out of them.

Parameters

• recordings (RecordingSet) – a RecordingSet manifest.

• supervisions (SupervisionSet) – a SupervisionSet manifest.

• output_dir (Union[Path, str]) – path where the Kaldi-style data directory will be created.

lhotse.kaldi.load_kaldi_text_mapping(path, must_exist=False)

Load Kaldi files such as utt2spk, spk2gender, text, etc. as a dict.

Return type

Dict[str, Optional[str]]

lhotse.kaldi.save_kaldi_text_mapping(data, path)

Save flat dicts to Kaldi files such as utt2spk, spk2gender, text, etc.

Others

Helper methods used throughout the codebase.

lhotse.manipulation.combine(*manifests)

Combine multiple manifests of the same type into one.

Examples:

>>> # Pass several arguments
>>> combine(recording_set1, recording_set2, recording_set3)
>>> # Or pass a single list/tuple of manifests
>>> combine([supervision_set1, supervision_set2])

Return type

~Manifest

lhotse.manipulation.to_manifest(items)

Take an iterable of data types in Lhotse such as Recording, SupervisonSegment or Cut, and create the manifest of the corresponding type. When the iterable is empty, returns None.

Return type

Optional[~Manifest]

lhotse.manipulation.load_manifest(path)

Generic utility for reading an arbitrary manifest.

Return type

~Manifest