custom_interface.py

import torch
from speechbrain.inference.interfaces import Pretrained

class AttentionMLP(torch.nn.Module):
    def __init__(self, input_dim, hidden_dim):
        super(AttentionMLP, self).__init__()
        self.layers = torch.nn.Sequential(
            torch.nn.Linear(input_dim, hidden_dim),
            torch.nn.ReLU(),
            torch.nn.Linear(hidden_dim, 1, bias=False),
        )

    def forward(self, x):
        x = self.layers(x)
        att_w = torch.nn.functional.softmax(x, dim=2)
        return att_w


class Discrete_EmbeddingLayer(torch.nn.Module):
    """This class handles embedding layers  for discrete tokens.

    Arguments
    ---------
    num_codebooks: int ,
        number of codebooks of the tokenizer.
    vocab_size : int,
        size of the dictionary of embeddings
    emb_dim: int ,
        the size of each embedding vector
    pad_index: int (default: 0),
        If specified, the entries at padding_idx do not contribute to the gradient.
    init: boolean (default: False):
        If set to True, init the embedding with the tokenizer embedding otherwise init randomly.
    freeze: boolean (default: False)
       If True, the embedding is frozen. If False, the model will be trained
        alongside with the rest of the pipeline.

    Example
    -------
    >>> from speechbrain.lobes.models.huggingface_transformers.encodec import Encodec
    >>> model_hub = "facebook/encodec_24khz"
    >>> save_path = "savedir"
    >>> model = Encodec(model_hub, save_path)
    >>> audio = torch.randn(4, 1000)
    >>> length = torch.tensor([1.0, .5, .75, 1.0])
    >>> tokens, emb = model.encode(audio, length)
    >>> print(tokens.shape)
    torch.Size([4, 4, 2])
    >>> emb= Discrete_EmbeddingLayer(2, 1024, 1024)
    >>> in_emb = emb(tokens)
    >>> print(in_emb.shape)
    torch.Size([4, 4, 2, 1024])
    """

    def __init__(
        self,
        num_codebooks,
        vocab_size,
        emb_dim,
        pad_index=0,
        init=False,
        freeze=False,
    ):
        super(Discrete_EmbeddingLayer, self).__init__()
        self.vocab_size = vocab_size
        self.num_codebooks = num_codebooks
        self.freeze = freeze
        self.embedding = torch.nn.Embedding(
            num_codebooks * vocab_size, emb_dim
        ).requires_grad_(not self.freeze)
        self.init = init

    def init_embedding(self, weights):
        with torch.no_grad():
            self.embedding.weight = torch.nn.Parameter(weights)

    def forward(self, in_tokens):
        """Computes the embedding for discrete tokens.
        a sample.

        Arguments
        ---------
        in_tokens : torch.Tensor
            A (Batch x Time x num_codebooks)
            audio sample
        Returns
        -------
        in_embs : torch.Tensor
        """
        with torch.set_grad_enabled(not self.freeze):
            #  Add unique token IDs across diffrent codebooks by adding num_codebooks * vocab_size
            in_tokens += torch.arange(
                0,
                self.num_codebooks * self.vocab_size,
                self.vocab_size,
                device=in_tokens.device,
            )
            # Forward Pass to embedding and
            in_embs = self.embedding(in_tokens)
            return in_embs

class CustomEncoderClassifier(Pretrained):
    """A ready-to-use class for utterance-level classification (e.g, speaker-id,
    language-id, emotion recognition, keyword spotting, etc).
    The class assumes that an self-supervised encoder like wav2vec2/hubert and a classifier model
    are defined in the yaml file. If you want to
    convert the predicted index into a corresponding text label, please
    provide the path of the label_encoder in a variable called 'lab_encoder_file'
    within the yaml.
    The class can be used either to run only the encoder (encode_batch()) to
    extract embeddings or to run a classification step (classify_batch()).
    ```
    Example
    -------
    >>> import torchaudio
    >>> from speechbrain.pretrained import EncoderClassifier
    >>> # Model is downloaded from the speechbrain HuggingFace repo
    >>> tmpdir = getfixture("tmpdir")
    >>> classifier = EncoderClassifier.from_hparams(
    ...     source="speechbrain/spkrec-ecapa-voxceleb",
    ...     savedir=tmpdir,
    ... )
    >>> # Compute embeddings
    >>> signal, fs = torchaudio.load("samples/audio_samples/example1.wav")
    >>> embeddings =  classifier.encode_batch(signal)
    >>> # Classification
    >>> prediction =  classifier .classify_batch(signal)
    """

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.similarity = torch.nn.CosineSimilarity(dim=-1, eps=1e-6)

    def encode_batch(self, wavs, wav_lens=None, normalize=False):
        """Encodes the input audio into a single vector embedding.
        The waveforms should already be in the model's desired format.
        You can call:
        ``normalized = <this>.normalizer(signal, sample_rate)``
        to get a correctly converted signal in most cases.
        Arguments
        ---------
        wavs : torch.tensor
            Batch of waveforms [batch, time, channels] or [batch, time]
            depending on the model. Make sure the sample rate is fs=16000 Hz.
        wav_lens : torch.tensor
            Lengths of the waveforms relative to the longest one in the
            batch, tensor of shape [batch]. The longest one should have
            relative length 1.0 and others len(waveform) / max_length.
            Used for ignoring padding.
        normalize : bool
            If True, it normalizes the embeddings with the statistics
            contained in mean_var_norm_emb.
        Returns
        -------
        torch.tensor
            The encoded batch
        """
        # Manage single waveforms in input
        if len(wavs.shape) == 1:
            wavs = wavs.unsqueeze(0)

        # Assign full length if wav_lens is not assigned
        if wav_lens is None:
            wav_lens = torch.ones(wavs.shape[0], device=self.device)

        # Storing waveform in the specified device
        wavs, wav_lens = wavs.to(self.device), wav_lens.to(self.device)
        wavs = wavs.float()

        with torch.no_grad():
            self.hparams.codec.to(self.device).eval()
            tokens, _, _ = self.hparams.codec(
                wavs, wav_lens, **self.hparams.tokenizer_config
            )
            embeddings = self.mods.discrete_embedding_layer(tokens)
            att_w = self.mods.attention_mlp(embeddings)
            feats = torch.matmul(att_w.transpose(2, -1), embeddings).squeeze(-2)
            embeddings = self.mods.embedding_model(feats, wav_lens)
        return embeddings.squeeze(1)


    def verify_batch(
        self, wavs1, wavs2, wav1_lens=None, wav2_lens=None, threshold=0.25
    ):
        """Performs speaker verification with cosine distance.

        It returns the score and the decision (0 different speakers,
        1 same speakers).

        Arguments
        ---------
        wavs1 : Torch.Tensor
            torch.Tensor containing the speech waveform1 (batch, time).
            Make sure the sample rate is fs=16000 Hz.
        wavs2 : Torch.Tensor
            torch.Tensor containing the speech waveform2 (batch, time).
            Make sure the sample rate is fs=16000 Hz.
        wav1_lens : Torch.Tensor
            torch.Tensor containing the relative length for each sentence
            in the length (e.g., [0.8 0.6 1.0])
        wav2_lens : Torch.Tensor
            torch.Tensor containing the relative length for each sentence
            in the length (e.g., [0.8 0.6 1.0])
        threshold : Float
            Threshold applied to the cosine distance to decide if the
            speaker is different (0) or the same (1).

        Returns
        -------
        score
            The score associated to the binary verification output
            (cosine distance).
        prediction
            The prediction is 1 if the two signals in input are from the same
            speaker and 0 otherwise.
        """
        emb1 = self.encode_batch(wavs1, wav1_lens, normalize=False)
        emb2 = self.encode_batch(wavs2, wav2_lens, normalize=False)
        score = self.similarity(emb1, emb2)
        return score, score > threshold

    def verify_files(self, path_x, path_y, **kwargs):
        """Speaker verification with cosine distance

        Returns the score and the decision (0 different speakers,
        1 same speakers).

        Arguments
        ---------
        path_x : str
            Path to file x
        path_y : str
            Path to file y
        **kwargs : dict
            Arguments to ``load_audio``

        Returns
        -------
        score
            The score associated to the binary verification output
            (cosine distance).
        prediction
            The prediction is 1 if the two signals in input are from the same
            speaker and 0 otherwise.
        """
        waveform_x = self.load_audio(path_x, **kwargs)
        waveform_y = self.load_audio(path_y, **kwargs)
        # Fake batches:
        batch_x = waveform_x.unsqueeze(0)
        batch_y = waveform_y.unsqueeze(0)
        # Verify:
        score, decision = self.verify_batch(batch_x, batch_y)
        # Squeeze:
        return score[0], decision[0]