MCPcopy
hub / github.com/Audio-AGI/AudioSep / wav_to_spectrogram_phase

Method wav_to_spectrogram_phase

models/base.py:91–113  ·  view source on GitHub ↗

Waveform to spectrogram. Args: input: (batch_size, segment_samples, channels_num) Outputs: output: (batch_size, channels_num, time_steps, freq_bins)

(self, input, eps=1e-10)

Source from the content-addressed store, hash-verified

89
90
91 def wav_to_spectrogram_phase(self, input, eps=1e-10):
92 """Waveform to spectrogram.
93
94 Args:
95 input: (batch_size, segment_samples, channels_num)
96
97 Outputs:
98 output: (batch_size, channels_num, time_steps, freq_bins)
99 """
100 sp_list = []
101 cos_list = []
102 sin_list = []
103 channels_num = input.shape[1]
104 for channel in range(channels_num):
105 mag, cos, sin = self.spectrogram_phase(input[:, channel, :], eps=eps)
106 sp_list.append(mag)
107 cos_list.append(cos)
108 sin_list.append(sin)
109
110 sps = torch.cat(sp_list, dim=1)
111 coss = torch.cat(cos_list, dim=1)
112 sins = torch.cat(sin_list, dim=1)
113 return sps, coss, sins
114
115 def wav_to_spectrogram(self, input, eps=0.):
116 """Waveform to spectrogram.

Callers 1

forwardMethod · 0.80

Calls 2

spectrogram_phaseMethod · 0.95
appendMethod · 0.80

Tested by

no test coverage detected