[Whisper] Add SpecAugment (#21298)

* Return and rescale attention_mask

* Add SpecAugment to Whisper modeling

* Fix test

* Update docstring

* Add SpecAug related parameters to model config

* Add the _mask_input_features function to doc

* Fix quality

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Remove dev comments

* Add test

* Resolve conflict

* feat: mask {feature, time} prob fast tests

* Apply suggestions from code review

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

---------

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: sanchit-gandhi <sanchit@huggingface.co>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

docs/source/en/model_doc/whisper.mdx

@@ -72,6 +72,7 @@ The original code can be found [here](https://github.com/openai/whisper).
 
 [[autodoc]] WhisperModel
     - forward
+    - _mask_input_features
 
 ## WhisperForConditionalGeneration
 

src/transformers/models/whisper/configuration_whisper.py

@@ -136,6 +136,35 @@ class WhisperConfig(PretrainedConfig):
         begin_suppress_tokens (`List[int]`, *optional*, defaults to `[220,50256]`):
             A list containing tokens that will be supressed at the beginning of the sampling process. Initialized as
             the token for `" "` (`blank_token_id`) and the `eos_token_id`
+        apply_spec_augment (`bool`, *optional*, defaults to `False`):
+            Whether to apply *SpecAugment* data augmentation to the outputs of the feature encoder. For reference see
+            [SpecAugment: A Simple Data Augmentation Method for Automatic Speech
+            Recognition](https://arxiv.org/abs/1904.08779).
+        mask_time_prob (`float`, *optional*, defaults to 0.05):
+            Percentage (between 0 and 1) of all feature vectors along the time axis which will be masked. The masking
+            procecure generates `mask_time_prob*len(time_axis)/mask_time_length` independent masks over the axis. If
+            reasoning from the propability of each feature vector to be chosen as the start of the vector span to be
+            masked, *mask_time_prob* should be `prob_vector_start*mask_time_length`. Note that overlap may decrease the
+            actual percentage of masked vectors. This is only relevant if `apply_spec_augment == True`.
+        mask_time_length (`int`, *optional*, defaults to 10):
+            Length of vector span along the time axis.
+        mask_time_min_masks (`int`, *optional*, defaults to 2),:
+            The minimum number of masks of length `mask_feature_length` generated along the time axis, each time step,
+            irrespectively of `mask_feature_prob`. Only relevant if ''mask_time_prob*len(time_axis)/mask_time_length <
+            mask_time_min_masks''
+        mask_feature_prob (`float`, *optional*, defaults to 0.0):
+            Percentage (between 0 and 1) of all feature vectors along the feature axis which will be masked. The
+            masking procecure generates `mask_feature_prob*len(feature_axis)/mask_time_length` independent masks over
+            the axis. If reasoning from the propability of each feature vector to be chosen as the start of the vector
+            span to be masked, *mask_feature_prob* should be `prob_vector_start*mask_feature_length`. Note that overlap
+            may decrease the actual percentage of masked vectors. This is only relevant if `apply_spec_augment is
+            True`.
+        mask_feature_length (`int`, *optional*, defaults to 10):
+            Length of vector span along the feature axis.
+        mask_feature_min_masks (`int`, *optional*, defaults to 0),:
+            The minimum number of masks of length `mask_feature_length` generated along the feature axis, each time
+            step, irrespectively of `mask_feature_prob`. Only relevant if
+            `mask_feature_prob*len(feature_axis)/mask_feature_length < mask_feature_min_masks`.
 
 
     Example:
@@ -185,6 +214,13 @@ def __init__(
         eos_token_id=50256,
         suppress_tokens=None,
         begin_suppress_tokens=[220, 50256],
+        apply_spec_augment=False,
+        mask_time_prob=0.05,
+        mask_time_length=10,
+        mask_time_min_masks=2,
+        mask_feature_prob=0.0,
+        mask_feature_length=10,
+        mask_feature_min_masks=0,
         **kwargs,
     ):
         self.vocab_size = vocab_size
@@ -208,6 +244,14 @@ def __init__(
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
         self.max_source_positions = max_source_positions
         self.max_target_positions = max_target_positions
+        # fine-tuning config parameters for SpecAugment: https://arxiv.org/abs/1904.08779
+        self.apply_spec_augment = apply_spec_augment
+        self.mask_time_prob = mask_time_prob
+        self.mask_time_length = mask_time_length
+        self.mask_time_min_masks = mask_time_min_masks
+        self.mask_feature_prob = mask_feature_prob
+        self.mask_feature_length = mask_feature_length
+        self.mask_feature_min_masks = mask_feature_min_masks
         super().__init__(
             pad_token_id=pad_token_id,
             bos_token_id=bos_token_id,

src/transformers/models/whisper/feature_extraction_whisper.py

@@ -307,6 +307,7 @@ def __call__(
             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,
         )
         # make sure list is in array format
         input_features = padded_inputs.get("input_features").transpose(2, 0, 1)
@@ -318,6 +319,10 @@ def __call__(
         else:
             padded_inputs["input_features"] = input_features
 
+        if return_attention_mask:
+            # rescale from sample (48000) to feature (3000)
+            padded_inputs["attention_mask"] = padded_inputs["attention_mask"][:, :: self.hop_length]
+
         if return_tensors is not None:
             padded_inputs = padded_inputs.convert_to_tensors(return_tensors)
 

src/transformers/models/whisper/modeling_whisper.py

@@ -19,6 +19,7 @@
 import random
 from typing import Optional, Tuple, Union
 
+import numpy as np
 import torch
 import torch.utils.checkpoint
 from torch import nn
@@ -97,6 +98,126 @@ def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int]
     return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
 
 
+# Copied from transformers.models.wav2vec2.modeling_wav2vec2._compute_mask_indices
+def _compute_mask_indices(
+    shape: Tuple[int, int],
+    mask_prob: float,
+    mask_length: int,
+    attention_mask: Optional[torch.LongTensor] = None,
+    min_masks: int = 0,
+) -> np.ndarray:
+    """
+    Computes random mask spans for a given shape. Used to implement [SpecAugment: A Simple Data Augmentation Method for
+    ASR](https://arxiv.org/abs/1904.08779). Note that this method is not optimized to run on TPU and should be run on
+    CPU as part of the preprocessing during training.
+
+    Args:
+        shape: The shape for which to compute masks. This should be of a tuple of size 2 where
+               the first element is the batch size and the second element is the length of the axis to span.
+        mask_prob:  The percentage of the whole axis (between 0 and 1) which will be masked. The number of
+                    independently generated mask spans of length `mask_length` is computed by
+                    `mask_prob*shape[1]/mask_length`. Note that due to overlaps, `mask_prob` is an upper bound and the
+                    actual percentage will be smaller.
+        mask_length: size of the mask
+        min_masks: minimum number of masked spans
+        attention_mask: A (right-padded) attention mask which independently shortens the feature axis of
+                        each batch dimension.
+    """
+    batch_size, sequence_length = shape
+
+    if mask_length < 1:
+        raise ValueError("`mask_length` has to be bigger than 0.")
+
+    if mask_length > sequence_length:
+        raise ValueError(
+            f"`mask_length` has to be smaller than `sequence_length`, but got `mask_length`: {mask_length}"
+            f" and `sequence_length`: {sequence_length}`"
+        )
+
+    # epsilon is used for probabilistic rounding
+    epsilon = np.random.rand(1).item()
+
+    def compute_num_masked_span(input_length):
+        """Given input length, compute how many spans should be masked"""
+        num_masked_span = int(mask_prob * input_length / mask_length + epsilon)
+        num_masked_span = max(num_masked_span, min_masks)
+
+        # make sure num masked span <= sequence_length
+        if num_masked_span * mask_length > sequence_length:
+            num_masked_span = sequence_length // mask_length
+
+        # make sure num_masked span is also <= input_length - (mask_length - 1)
+        if input_length - (mask_length - 1) < num_masked_span:
+            num_masked_span = max(input_length - (mask_length - 1), 0)
+
+        return num_masked_span
+
+    # compute number of masked spans in batch
+    input_lengths = (
+        attention_mask.sum(-1).detach().tolist()
+        if attention_mask is not None
+        else [sequence_length for _ in range(batch_size)]
+    )
+
+    # SpecAugment mask to fill
+    spec_aug_mask = np.zeros((batch_size, sequence_length), dtype=bool)
+    spec_aug_mask_idxs = []
+
+    max_num_masked_span = compute_num_masked_span(sequence_length)
+
+    if max_num_masked_span == 0:
+        return spec_aug_mask
+
+    for input_length in input_lengths:
+        # compute num of masked spans for this input
+        num_masked_span = compute_num_masked_span(input_length)
+
+        # get random indices to mask
+        spec_aug_mask_idx = np.random.choice(
+            np.arange(input_length - (mask_length - 1)), num_masked_span, replace=False
+        )
+
+        # pick first sampled index that will serve as a dummy index to pad vector
+        # to ensure same dimension for all batches due to probabilistic rounding
+        # Picking first sample just pads those vectors twice.
+        if len(spec_aug_mask_idx) == 0:
+            # this case can only happen if `input_length` is strictly smaller then
+            # `sequence_length` in which case the last token has to be a padding
+            # token which we can use as a dummy mask id
+            dummy_mask_idx = sequence_length - 1
+        else:
+            dummy_mask_idx = spec_aug_mask_idx[0]
+
+        spec_aug_mask_idx = np.concatenate(
+            [spec_aug_mask_idx, np.ones(max_num_masked_span - num_masked_span, dtype=np.int32) * dummy_mask_idx]
+        )
+        spec_aug_mask_idxs.append(spec_aug_mask_idx)
+
+    spec_aug_mask_idxs = np.array(spec_aug_mask_idxs)
+
+    # expand masked indices to masked spans
+    spec_aug_mask_idxs = np.broadcast_to(
+        spec_aug_mask_idxs[:, :, None], (batch_size, max_num_masked_span, mask_length)
+    )
+    spec_aug_mask_idxs = spec_aug_mask_idxs.reshape(batch_size, max_num_masked_span * mask_length)
+
+    # add offset to the starting indexes so that indexes now create a span
+    offsets = np.arange(mask_length)[None, None, :]
+    offsets = np.broadcast_to(offsets, (batch_size, max_num_masked_span, mask_length)).reshape(
+        batch_size, max_num_masked_span * mask_length
+    )
+    spec_aug_mask_idxs = spec_aug_mask_idxs + offsets
+
+    # ensure that we cannot have indices larger than sequence_length
+    if spec_aug_mask_idxs.max() > sequence_length - 1:
+        spec_aug_mask_idxs[spec_aug_mask_idxs > sequence_length - 1] = sequence_length - 1
+
+    # scatter indices to mask
+    np.put_along_axis(spec_aug_mask, spec_aug_mask_idxs, 1, -1)
+
+    return spec_aug_mask
+
+
 class WhisperPositionalEmbedding(nn.Embedding):
     def __init__(self, num_positions: int, embedding_dim: int, padding_idx: Optional[int] = None):
         super().__init__(num_positions, embedding_dim)
@@ -503,6 +624,14 @@ def _get_feat_extract_output_lengths(self, input_lengths: torch.LongTensor):
             the soundfile library (`pip install soundfile`). To prepare the array into `input_features`, the
             [`AutoFeatureExtractor`] should be used for extracting the mel features, padding and conversion into a
             tensor of type `torch.FloatTensor`. See [`~WhisperFeatureExtractor.__call__`]
+        attention_mask (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Mask to avoid performing *SpecAugment* data augmentation on padding token indices. Mask values selected in
+            `[0, 1]`:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            [What are attention masks?](../glossary#attention-mask)
         decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
             Indices of decoder input sequence tokens in the vocabulary.
 
@@ -999,11 +1128,55 @@ def freeze_encoder(self):
         """
         self.encoder._freeze_parameters()
 
+    def _mask_input_features(
+        self,
+        input_features: torch.FloatTensor,
+        attention_mask: Optional[torch.LongTensor] = None,
+    ):
+        """
+        Masks extracted features along time axis and/or along feature axis according to
+        [SpecAugment](https://arxiv.org/abs/1904.08779).
+        """
+
+        # `config.apply_spec_augment` can set masking to False
+        if not getattr(self.config, "apply_spec_augment", True):
+            return input_features
+
+        # generate indices & apply SpecAugment along time axis
+        batch_size, hidden_size, sequence_length = input_features.size()
+
+        if self.config.mask_time_prob > 0 and self.training:
+            # generate indices & apply SpecAugment along time axis
+            mask_time_indices = _compute_mask_indices(
+                (batch_size, sequence_length),
+                mask_prob=self.config.mask_time_prob,
+                mask_length=self.config.mask_time_length,
+                attention_mask=attention_mask,
+                min_masks=self.config.mask_time_min_masks,
+            )
+            mask_time_indices = torch.tensor(mask_time_indices, device=input_features.device, dtype=torch.bool)
+            mask_time_indices = mask_time_indices[:, None].expand(-1, hidden_size, -1)
+            input_features[mask_time_indices] = 0
+
+        if self.config.mask_feature_prob > 0 and self.training:
+            # generate indices & apply SpecAugment along feature axis
+            mask_feature_indices = _compute_mask_indices(
+                (batch_size, hidden_size),
+                mask_prob=self.config.mask_feature_prob,
+                mask_length=self.config.mask_feature_length,
+                min_masks=self.config.mask_feature_min_masks,
+            )
+            mask_feature_indices = torch.tensor(mask_feature_indices, device=input_features.device, dtype=torch.bool)
+            input_features[mask_feature_indices] = 0
+
+        return input_features
+
     @add_start_docstrings_to_model_forward(WHISPER_INPUTS_DOCSTRING)
     @replace_return_docstrings(output_type=Seq2SeqModelOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
-        input_features: Optional[torch.LongTensor] = None,
+        input_features: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.LongTensor] = None,
         decoder_input_ids: Optional[torch.LongTensor] = None,
         decoder_attention_mask: Optional[torch.LongTensor] = None,
         head_mask: Optional[torch.Tensor] = None,
@@ -1044,6 +1217,8 @@ def forward(
         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
 
         if encoder_outputs is None:
+            input_features = self._mask_input_features(input_features, attention_mask=attention_mask)
+
             encoder_outputs = self.encoder(
                 input_features,
                 head_mask=head_mask,
@@ -1139,7 +1314,8 @@ def freeze_encoder(self):
     @replace_return_docstrings(output_type=Seq2SeqLMOutput, config_class=_CONFIG_FOR_DOC)
     def forward(
         self,
-        input_features: Optional[torch.LongTensor] = None,
+        input_features: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.LongTensor] = None,
         decoder_input_ids: Optional[torch.LongTensor] = None,
         decoder_attention_mask: Optional[torch.LongTensor] = None,
         head_mask: Optional[torch.Tensor] = None,
@@ -1193,6 +1369,7 @@ def forward(
 
         outputs = self.model(
             input_features,
+            attention_mask=attention_mask,
             decoder_input_ids=decoder_input_ids,
             encoder_outputs=encoder_outputs,
             decoder_attention_mask=decoder_attention_mask,