decoder的feature_extractors 输入的张量是(batch_size, length)吗

[HigginsGurry]

feature_extractors 输入的张量是(batch_size, length)吗，audio=audio.unsqueeze(1)补全成(batch_size, channel, length)?

[jishengpeng]

feature_extractors 输入的张量是(batch_size, length)吗，audio=audio.unsqueeze(1)补全成(batch_size, channel, length)?

yes, length is original audio. But the feature_extractor is the encoder not decoder.