Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/calclavia/tal-asrd
/ functions
Functions
314 in github.com/calclavia/tal-asrd
⨍
Functions
314
◇
Types & classes
35
↓ 1 callers
Function
download_model
Download pretrained wav2vec Args: loc (str): Save location for model url (str): Download URI overwrite (bool, op
tal/diarization/features/wav2vec/__init__.py:18
↓ 1 callers
Method
encode
Args: x - Tensor of raw waveform [batch, length]
tal/asr/models.py:195
↓ 1 callers
Method
encode_features
(self, x: torch.Tensor, audio_lens: torch.LongTensor = None)
tal/asr/models.py:164
↓ 1 callers
Method
encode_features
(self, x: torch.Tensor, audio_lens: torch.LongTensor = None)
tal/asr/models.py:440
↓ 1 callers
Function
enforce_cluster_id_uniqueness
Enforce uniqueness of cluster id across sequences. Args: cluster_ids: a list of 1-dim list/numpy.ndarray of strings Returns:
tal/diarization/uisrnn/utils.py:46
↓ 1 callers
Method
extract
Feature extraction network. RF: 21 frames (0.21s)
tal/asr/models.py:374
↓ 1 callers
Method
extract_features
(self, x, specaug=True)
tal/asr/models.py:154
↓ 1 callers
Method
extract_features
(self, x, specaug=True)
tal/asr/models.py:430
↓ 1 callers
Method
fit_concatenated
Fit UISRNN model to concatenated sequence and cluster_id. Args: train_sequence: the training observation sequence, which is a
tal/diarization/uisrnn/uisrnn.py:159
↓ 1 callers
Method
forward
(self, x)
tal/asr/train_embed.py:43
↓ 1 callers
Function
frame_generator
Generates audio frames from PCM audio data. Takes the desired frame duration in milliseconds, the PCM data, and the sample rate. Yields
tal/vad/webrtcvad.py:58
↓ 1 callers
Function
generate_random_string
Generate a random string of upper case letters and digits. Args: length: length of the generated string Returns: the generat
tal/diarization/uisrnn/utils.py:32
↓ 1 callers
Method
generate_single
(self, batch)
tal/asr/system.py:586
↓ 1 callers
Method
generate_single
(self, batch)
tal/baseline/speaker_system.py:107
↓ 1 callers
Method
generate_unaligned
Generates text from encoder features. Args: audio_x: Input audio (wav). [batch, seq len] generated: The i
tal/asr/system.py:254
↓ 1 callers
Function
get_audio_info_wav
Get audio information for a WAV file Args: audio_loc (str): Target audio file path Returns: str: Codec type int
tal/utils/audio.py:130
↓ 1 callers
Function
get_speaker_frames
(ep_turns, file_duration)
tal/baseline/reconcile.py:51
↓ 1 callers
Function
get_speaker_ids
(wav_loc, model)
tal/baseline/reconcile.py:76
↓ 1 callers
Function
get_speech_frames
( audio_segment: np.array, frame_duration_ms: int = 30, sample_rate: int = 16000, vad_leve
tal/asr/speech_detect.py:44
↓ 1 callers
Function
get_wder_edits
Given a list of reference words and hypothesis words, find the edit operations (substitution and no-ops only) between the two.
tal/wder_search_emb.py:250
↓ 1 callers
Function
get_wder_edits
Given a list of reference words and hypothesis words, find the edit operations (substitution and no-ops only) between the two.
tal/wder_search_emb_new_format.py:239
↓ 1 callers
Function
get_wder_edits
Given a list of reference words and hypothesis words, find the edit operations (substitution and no-ops only) between the two.
tal/wder_search.py:225
↓ 1 callers
Function
is_valid_segment
Check if this list of utterances contain valid audio timestamps.
tal/asr/data/aligned.py:64
↓ 1 callers
Function
is_valid_segment
Check if this list of utterances contain valid audio timestamps.
tal/asr/data/baseline_speaker.py:63
↓ 1 callers
Function
load_model
(args: object = None, device: str = DEVICE)
tal/asr/transcribe.py:237
↓ 1 callers
Function
main
(args)
tal/asr/speech_detect.py:182
↓ 1 callers
Function
main
The main function.
tal/diarization/uisrnn/train.py:267
↓ 1 callers
Function
main
The main function.
tal/diarization/uisrnn/train_large.py:270
↓ 1 callers
Function
ngram_repeat_mask
(xs, n)
tal/asr/util.py:5
↓ 1 callers
Function
overlap_ix
(a: str, b: str, word_overlap: int = 5)
tal/asr/transcribe.py:29
↓ 1 callers
Function
prune_bad_utterances
Prunes away "bad utterances" by thresholding all utterances above a certain loss threshold, then removing them from index. Args:
tal/utils/prune_bad_utterances.py:6
↓ 1 callers
Function
read_wave
Reads a .wav file. Takes the path, and returns (PCM audio data, sample rate).
tal/asr/speech_detect.py:10
↓ 1 callers
Function
read_wave
Reads a .wav file. Takes the path, and returns (PCM audio data, sample rate).
tal/vad/webrtcvad.py:23
↓ 1 callers
Function
run_experiment
Run a single experiment Args: train_f (list): Sequence of files, each containing a T x H array (audio features) train_cluste
tal/diarization/uisrnn/train_large.py:22
↓ 1 callers
Function
run_vad
( aggressiveness: int, file_loc: str, raw_audio: bytes = None, sample_rate
tal/vad/webrtcvad.py:145
↓ 1 callers
Function
sample_permuted_segments
Sample sequences with permuted blocks. Args: index_sequence: (integer array, size: L) - subsequence index For exa
tal/diarization/uisrnn/utils.py:139
↓ 1 callers
Method
state_dict
(self)
tal/optimizers.py:270
↓ 1 callers
Function
tokenize
(stuff)
tal/alignment/aeneas.py:26
↓ 1 callers
Function
tokenize_utterances_word_align
( utterances: list, filtered_utts: list, start_time: float, end_time: float, segment_
tal/asr/data/util.py:125
↓ 1 callers
Method
training_step
(self, batch, batch_idx)
tal/asr/train_embed.py:46
↓ 1 callers
Method
training_step
(self, batch, batch_nb)
tal/asr/system.py:529
↓ 1 callers
Method
training_step
(self, batch, batch_nb)
tal/baseline/speaker_system.py:65
↓ 1 callers
Function
transcribe_batch
(batch: list, model: nn.Module, beam_width: int = 4,
tal/asr/transcribe.py:172
↓ 1 callers
Function
transcribe_file
( audio_path: str, model: nn.Module, window_frames: int, stride_frames: in
tal/asr/transcribe.py:79
↓ 1 callers
Function
tweet_tokenize
(stuff)
tal/asr/data/util.py:10
↓ 1 callers
Function
vad_collector
Filters out non-voiced audio frames. Given a webrtcvad.Vad and a source of audio frames, yields only the voiced audio. Uses a padded, s
tal/vad/webrtcvad.py:74
↓ 1 callers
Method
validation_step
(self, batch, batch_idx)
tal/asr/train_embed.py:73
↓ 1 callers
Function
write_wave
Writes a .wav file. Takes path, PCM audio data, and sample rate.
tal/asr/speech_detect.py:25
Method
__call__
Creates a batch out of samples
tal/asr/data/segment.py:310
Method
__call__
Creates a batch out of samples
tal/asr/data/aligned.py:246
Method
__call__
Creates a batch out of samples
tal/asr/data/audio.py:110
Method
__call__
Creates a batch out of samples
tal/asr/data/baseline_speaker.py:220
Method
__enter__
(self)
tal/__init__.py:44
Method
__exit__
(self, exc_type, exc_val, exc_tb)
tal/__init__.py:48
Method
__getitem__
(self, i: int)
tal/asr/data/segment.py:177
Method
__getitem__
(self, i: int)
tal/asr/data/aligned.py:199
Method
__getitem__
(self, i: int)
tal/asr/data/audio.py:75
Method
__getitem__
(self, i: int)
tal/asr/data/baseline_speaker.py:186
Method
__getstate__
(self)
tal/asr/logger.py:17
Method
__getstate__
Pickle serialization
tal/asr/tokenizers/sentencepiece.py:35
Method
__init__
(self, d_model, dropout=0.1, max_len=5000)
tal/modules.py:41
Method
__init__
(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=0)
tal/optimizers.py:11
Method
__init__
(self, params, lr=None, eps=(1e-30, 1e-3), clip_threshold=1.0, decay_rate=-0.8, beta1=None, w
tal/optimizers.py:113
Method
__init__
(self, optimizer, k=5, alpha=0.5)
tal/optimizers.py:236
Method
__init__
( self, weights_path: str = '/root/data4/bernard-asr-models/tal-tune2/a
tal/asr/transcribe.py:214
Method
__init__
(self, args)
tal/asr/train_embed.py:22
Method
__init__
(self, sr=DEFAULT_SR, n_mels=80, eps=1e-6)
tal/asr/models.py:22
Method
__init__
(self, model_type='2x', num_speakers=0, n_mels=80,
tal/asr/models.py:57
Method
__init__
(self, hidden, kernel_size, groups, dropout=0.1)
tal/asr/models.py:299
Method
__init__
(self, num_speakers=6008, n_mels=80, dropout=0.2,
tal/asr/models.py:401
Method
__init__
(self, d_model, nhead, dim_feedforward=2048, dropout=0.1, activation="relu")
tal/asr/models.py:491
Method
__init__
(self, **kwargs)
tal/asr/logger.py:8
Method
__init__
(self, smoothing=0.0)
tal/asr/util.py:20
Method
__init__
(self, bytes, timestamp, duration)
tal/asr/speech_detect.py:38
Method
__init__
(self, args)
tal/asr/system.py:27
Method
__init__
(self, cache_path: str = None, **kwargs)
tal/asr/tokenizers/__init__.py:8
Method
__init__
(self, cache_path: str = None, **kwargs)
tal/asr/tokenizers/sentencepiece.py:18
Method
__init__
(self, cache_path: str = None, **kwargs)
tal/asr/tokenizers/transformers.py:12
Method
__init__
Args: data_dir (str): Path to directory containing data files. Data folder is expected to contain
tal/asr/data/segment.py:81
Method
__init__
(self, padding_idx)
tal/asr/data/segment.py:307
Method
__init__
Args: data_dir (str): Path to directory containing data files. Data folder is expected to contain
tal/asr/data/aligned.py:97
Method
__init__
(self, padding_idx)
tal/asr/data/aligned.py:243
Method
__init__
Args: data_dir: Path to dataset max_duration: Number of seconds for an audio segment ext: File extens
tal/asr/data/audio.py:34
Method
__init__
Args: data_dir (str): Path to directory containing data files. Data folder is expected to contain a '
tal/asr/data/baseline_speaker.py:96
Method
__init__
(self, padding_idx)
tal/asr/data/baseline_speaker.py:217
Method
__init__
(self, input_dim, hidden_size, depth, observation_dim, dropout=0)
tal/diarization/uisrnn/uisrnn.py:23
Method
__init__
(self, source=None)
tal/diarization/uisrnn/uisrnn.py:45
Method
__init__
(self, verbosity)
tal/diarization/uisrnn/utils.py:18
Method
__init__
(self, bytes, timestamp, duration)
tal/vad/webrtcvad.py:52
Method
__init__
(self, args)
tal/baseline/speaker_system.py:27
Method
__len__
(self)
tal/asr/tokenizers/__init__.py:19
Method
__len__
(self)
tal/asr/tokenizers/sentencepiece.py:32
Method
__len__
(self)
tal/asr/tokenizers/transformers.py:33
Method
__len__
(self)
tal/asr/data/segment.py:174
Method
__len__
(self)
tal/asr/data/aligned.py:196
Method
__len__
(self)
tal/asr/data/audio.py:72
Method
__len__
(self)
tal/asr/data/baseline_speaker.py:183
Method
__setstate__
(self, state)
tal/optimizers.py:16
Method
__setstate__
Pickle deserialization
tal/asr/tokenizers/sentencepiece.py:41
Method
_encode
Encodes a sentence to token IDs Args: sentence (str): Sentence in text form Returns: Sequenc
tal/asr/tokenizers/sentencepiece.py:47
← previous
next →
101–200 of 314, ranked by callers