File size: 11,039 Bytes
ea174b0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 |
import numpy as np
import torch
from typing import Union, List, Optional
from transformers.feature_extraction_sequence_utils import SequenceFeatureExtractor
from transformers.feature_extraction_utils import BatchFeature
from transformers.utils import TensorType, logging
from transformers.utils.import_utils import is_torch_available
from transformers.audio_utils import mel_filter_bank, spectrogram, window_function
class MelFeatureExtractor(SequenceFeatureExtractor):
model_input_names = ["input_features"]
def __init__(
self,
feature_size=80,
sampling_rate=16000,
hop_length=160,
chunk_length=30,
n_fft=400,
padding_value=0.0,
dither=0.0,
return_attention_mask=False,
max_frequency=None,
**kwargs,
):
super().__init__(
feature_size=feature_size,
sampling_rate=sampling_rate,
padding_value=padding_value,
return_attention_mask=return_attention_mask,
**kwargs,
)
self.n_fft = n_fft
self.hop_length = hop_length
self.chunk_length = chunk_length
self.n_samples = chunk_length * sampling_rate
self.nb_max_frames = self.n_samples // hop_length
self.sampling_rate = sampling_rate
self.dither = dither
self.max_frequency = max_frequency if max_frequency is not None else sampling_rate / 2
self.mel_filters = mel_filter_bank(
num_frequency_bins=1 + n_fft // 2,
num_mel_filters=feature_size,
min_frequency=0.0,
max_frequency=self.max_frequency,
sampling_rate=sampling_rate,
norm="slaney",
mel_scale="slaney",
)
def _np_extract_fbank_features(self, waveform_batch: np.array, device: str) -> np.ndarray:
if device != "cpu":
raise ValueError(
f"Got device `{device}` for feature extraction, but feature extraction on CUDA accelerator "
"devices requires torch, which is not installed. Either set `device='cpu'`, or "
"install torch according to the official instructions: https://pytorch.org/get-started/locally/"
)
log_spec_batch = []
for waveform in waveform_batch:
log_spec = spectrogram(
waveform,
window_function(self.n_fft, "hann"),
frame_length=self.n_fft,
hop_length=self.hop_length,
power=2.0,
dither=self.dither,
mel_filters=self.mel_filters,
log_mel="log10",
)
log_spec = log_spec[:, :-1]
log_spec = np.maximum(log_spec, log_spec.max() - 8.0)
log_spec = (log_spec + 4.0) / 4.0
log_spec_batch.append(log_spec)
log_spec_batch = np.array(log_spec_batch)
return log_spec_batch
def _torch_extract_fbank_features(self, waveform: np.array, device: str = "cpu") -> np.ndarray:
"""
Compute the log-mel spectrogram of the audio using PyTorch's GPU-accelerated STFT implementation with batching,
yielding results similar to cpu computing with 1e-5 tolerance.
"""
waveform = torch.from_numpy(waveform).to(device, torch.float32)
window = torch.hann_window(self.n_fft, device=device)
if self.dither != 0.0:
waveform += self.dither * torch.randn(waveform.shape, dtype=waveform.dtype, device=waveform.device)
stft = torch.stft(waveform, self.n_fft, self.hop_length, window=window, return_complex=True)
magnitudes = stft[..., :-1].abs() ** 2
mel_filters = torch.from_numpy(self.mel_filters).to(device, torch.float32)
mel_spec = mel_filters.T @ magnitudes
log_spec = torch.clamp(mel_spec, min=1e-10).log10()
if waveform.dim() == 2:
max_val = log_spec.max(dim=2, keepdim=True)[0].max(dim=1, keepdim=True)[0]
log_spec = torch.maximum(log_spec, max_val - 8.0)
else:
log_spec = torch.maximum(log_spec, log_spec.max() - 8.0)
log_spec = (log_spec + 4.0) / 4.0
if device != "cpu":
log_spec = log_spec.detach().cpu()
return log_spec.numpy()
@staticmethod
def zero_mean_unit_var_norm(
input_values: List[np.ndarray], attention_mask: List[np.ndarray], padding_value: float = 0.0
) -> List[np.ndarray]:
"""
Every array in the list is normalized to have zero mean and unit variance
"""
if attention_mask is not None:
attention_mask = np.array(attention_mask, np.int32)
normed_input_values = []
for vector, length in zip(input_values, attention_mask.sum(-1)):
normed_slice = (vector - vector[:length].mean()) / np.sqrt(vector[:length].var() + 1e-7)
if length < normed_slice.shape[0]:
normed_slice[length:] = padding_value
normed_input_values.append(normed_slice)
else:
normed_input_values = [(x - x.mean()) / np.sqrt(x.var() + 1e-7) for x in input_values]
return normed_input_values
def __call__(
self,
raw_speech: Union[np.ndarray, List[float], List[np.ndarray], List[List[float]]],
truncation: bool = True,
pad_to_multiple_of: Optional[int] = None,
return_tensors: Optional[Union[str, TensorType]] = None,
return_attention_mask: Optional[bool] = None,
padding: Optional[str] = "max_length",
max_length: Optional[int] = None,
sampling_rate: Optional[int] = None,
do_normalize: Optional[bool] = None,
device: Optional[str] = "cpu",
return_token_timestamps: Optional[bool] = None,
**kwargs,
) -> BatchFeature:
"""
Main method to featurize and prepare for the model one or several sequence(s). Implementation uses PyTorch for
the STFT computation if available, otherwise a slower NumPy based one.
Args:
raw_speech (`np.ndarray`, `List[float]`, `List[np.ndarray]`, `List[List[float]]`):
The sequence or batch of sequences to be padded. Each sequence can be a numpy array, a list of float
values, a list of numpy arrays or a list of list of float values. Must be mono channel audio, not
stereo, i.e. single float per timestep.
truncation (`bool`, *optional*, default to `True`):
Activates truncation to cut input sequences longer than *max_length* to *max_length*.
pad_to_multiple_of (`int`, *optional*, defaults to None):
If set will pad the sequence to a multiple of the provided value.
return_tensors (`str` or [`~utils.TensorType`], *optional*):
If set, will return tensors instead of list of python integers. Acceptable values are:
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return Numpy `np.ndarray` objects.
sampling_rate (`int`, *optional*):
The sampling rate at which the `raw_speech` input was sampled. If provided, it is checked against
the extractor's sampling rate.
padding_value (`float`, *optional*, defaults to 0.0):
The value that is used to fill the padding values / vectors.
do_normalize (`bool`, *optional*, defaults to `False`):
Whether or not to zero-mean unit-variance normalize the input.
device (`str`, *optional*, defaults to `'cpu'`):
Specifies the device for computation of the log-mel spectrogram.
return_token_timestamps (`bool`, *optional*, defaults to `None`):
Whether or not to return the number of frames of the input raw_speech.
"""
if sampling_rate is not None and sampling_rate != self.sampling_rate:
logger.warning(
f"The provided `raw_speech` input was sampled at {sampling_rate}Hz, but the feature extractor "
f"is configured for {self.sampling_rate}Hz. You should resample the audio to match the "
f"extractor's sampling rate to ensure correct feature extraction."
)
is_batched_numpy = isinstance(raw_speech, np.ndarray) and len(raw_speech.shape) > 1
if is_batched_numpy and len(raw_speech.shape) > 2:
raise ValueError(f"Only mono-channel audio is supported for input to {self}")
is_batched = is_batched_numpy or (
isinstance(raw_speech, (list, tuple)) and (isinstance(raw_speech[0], (np.ndarray, tuple, list)))
)
if is_batched:
raw_speech = [np.asarray([speech], dtype=np.float32).T for speech in raw_speech]
elif not is_batched and not isinstance(raw_speech, np.ndarray):
raw_speech = np.asarray(raw_speech, dtype=np.float32)
elif isinstance(raw_speech, np.ndarray) and raw_speech.dtype is np.dtype(np.float64):
raw_speech = raw_speech.astype(np.float32)
if not is_batched:
raw_speech = [np.asarray([raw_speech]).T]
batched_speech = BatchFeature({"input_features": raw_speech})
padded_inputs = self.pad(
batched_speech,
padding=padding,
max_length=max_length if max_length else self.n_samples,
truncation=truncation,
pad_to_multiple_of=pad_to_multiple_of,
return_attention_mask=return_attention_mask or do_normalize,
)
if do_normalize:
padded_inputs["input_features"] = self.zero_mean_unit_var_norm(
padded_inputs["input_features"],
attention_mask=padded_inputs["attention_mask"],
padding_value=self.padding_value,
)
padded_inputs["input_features"] = np.stack(padded_inputs["input_features"], axis=0)
input_features = padded_inputs.get("input_features").transpose(2, 0, 1)
extract_fbank_features = (
self._torch_extract_fbank_features if is_torch_available() else self._np_extract_fbank_features
)
input_features = extract_fbank_features(input_features[0], device)
if isinstance(input_features[0], List):
padded_inputs["input_features"] = [np.asarray(feature, dtype=np.float32) for feature in input_features]
else:
padded_inputs["input_features"] = input_features
if return_attention_mask:
padded_inputs["attention_mask"] = padded_inputs["attention_mask"][:, :: self.hop_length]
if return_token_timestamps is not None:
padded_inputs["num_frames"] = [len(raw_speech_i) // self.hop_length for raw_speech_i in raw_speech]
if return_tensors is not None:
padded_inputs = padded_inputs.convert_to_tensors(return_tensors)
return padded_inputs |