funasr/models/e2e_uni_asr.py
@@ -206,7 +206,7 @@ with torch.no_grad(): speech_raw, encoder_out, encoder_out_lens = self.encode(speech, speech_lengths, ind=ind) else: speech_raw, encoder_out_lens = self.encode(speech, speech_lengths, ind=ind) speech_raw, encoder_out, encoder_out_lens = self.encode(speech, speech_lengths, ind=ind) intermediate_outs = None if isinstance(encoder_out, tuple):
