Remove torchaudio

Signed-off-by: Jael Gu <mengjia.gu@zilliz.com>

torch_vggish.py

@@ -54,7 +54,7 @@ class Vggish(NNOperator):
         self.model.to(self.device)
 
     def __call__(self, datas: List[NamedTuple('data', [('audio', 'ndarray'), ('sample_rate', 'int')])]) -> numpy.ndarray:
-        audios = numpy.stack([item.audio for item in datas])
+        audios = numpy.hstack([item.audio for item in datas])
         sr = datas[0].sample_rate
         audio_array = numpy.reshape(audios, (-1, 1))
         audio_tensors = self.preprocess(audio_array, sr).to(self.device)

vggish_input.py

@@ -23,8 +23,6 @@ import resampy
 import mel_features
 import vggish_params
 
-import torchaudio
-
 
 def waveform_to_examples(data, sample_rate, return_tensor=True):
     """Converts audio waveform into an array of examples for VGGish.
@@ -80,17 +78,3 @@ def waveform_to_examples(data, sample_rate, return_tensor=True):
     return log_mel_examples
 
 
-def wavfile_to_examples(wav_file, return_tensor=True):
-    """Convenience wrapper around waveform_to_examples() for a common WAV format.
-
-  Args:
-    wav_file: String path to a file, or a file-like object. The file
-    is assumed to contain WAV audio data with signed 16-bit PCM samples.
-    torch: Return data as a Pytorch tensor ready for VGGish
-
-  Returns:
-    See waveform_to_examples.
-  """
-    data, sr = torchaudio.load(wav_file)
-    wav_data = data.detach().numpy().transpose()
-    return waveform_to_examples(wav_data, sr, return_tensor)