Function extract_fbank

funasr/utils/load_utils.py:277–315

Extract filter-bank features from audio data. Args: data: Audio samples (list of numpy arrays or tensors). data_len: Lengths of each sample. data_type (str): Input type ("sound", "fbank"). frontend: Frontend instance for feature extraction. Returns:

(data, data_len=None, data_type: str = "sound", frontend=None, **kwargs)

def extract_fbank(data, data_len=None, data_type: str = "sound", frontend=None, **kwargs):
    """Extract filter-bank features from audio data.

    Args:
        data: Audio samples (list of numpy arrays or tensors).
        data_len: Lengths of each sample.
        data_type (str): Input type ("sound", "fbank").
        frontend: Frontend instance for feature extraction.

    Returns:
        tuple: (features_tensor, feature_lengths, feature_times)
    """
    if isinstance(data, np.ndarray):
        data = torch.from_numpy(data)
        if len(data.shape) < 2:
            data = data[None, :]  # data: [batch, N]
        elif data.shape[0] > 1:
            data = data.mean(dim=0, keepdim=True)  # convert stereo/multi-channel to mono
        data_len = [data.shape[1]] if data_len is None else data_len
    elif isinstance(data, torch.Tensor):
        if len(data.shape) < 2:
            data = data[None, :]  # data: [batch, N]
        elif data.shape[0] > 1:
            data = data.mean(dim=0, keepdim=True)  # convert stereo/multi-channel to mono
        data_len = [data.shape[1]] if data_len is None else data_len
    elif isinstance(data, (list, tuple)):
        data_list, data_len = [], []
        for data_i in data:
            if isinstance(data_i, np.ndarray):
                data_i = torch.from_numpy(data_i)
            data_list.append(data_i)
            data_len.append(data_i.shape[0])
        data = pad_sequence(data_list, batch_first=True)  # data: [batch, N]

    data, data_len = frontend(data, data_len, **kwargs)

    if isinstance(data_len, (list, tuple)):
        data_len = torch.tensor([data_len])
    return data.to(torch.float32), data_len.to(torch.int32)


def _load_audio_ffmpeg(file: str, sr: int = 16000):