志浩
2023-04-07 0e3b3e0482b1117738b3d45d0fe9078ec963159e
funasr/bin/asr_inference_mfcca.py
@@ -194,8 +194,8 @@
        # Input as audio signal
        if isinstance(speech, np.ndarray):
            speech = torch.tensor(speech)
        if(speech.dim()==3):
            speech = torch.squeeze(speech, 2)
        #speech = speech.unsqueeze(0).to(getattr(torch, self.dtype))
        speech = speech.to(getattr(torch, self.dtype))
        # lenghts: (1,)