ValueError 从 Wav2Vec2 获取发射 PyTorch 管道 Model

Question

When calling

model = torchaudio.pipelines.WAV2VEC2_ASR_BASE_960H.get_model()
emission = model(data)

This is to get the emission probabilities from the model.

but I get

File "XXX\lib\site-packages\torch\nn\modules\module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "XXX\lib\site-packages\torchaudio\models\wav2vec2\model.py", line 119, in forward
    x, lengths = self.feature_extractor(waveforms, lengths)
  File "XXX\lib\site-packages\torch\nn\modules\module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "XXX\lib\site-packages\torchaudio\models\wav2vec2\components.py", line 135, in forward
    raise ValueError("Expected the input Tensor to be 2D (batch, time), " "but received {list(x.shape)}")
ValueError: Expected the input Tensor to be 2D (batch, time), but received {list(x.shape)}

Answer 1

what is the shape of data ?

It's probably you are feeding 1D tensor. The shape of the tensor has to be (batch, time).