funasr/models/seaco_paraformer/model.py
@@ -337,7 +337,8 @@ meta_data[ "batch_data_time"] = speech_lengths.sum().item() * frontend.frame_shift * frontend.lfr_n / 1000 speech.to(device=kwargs["device"]), speech_lengths.to(device=kwargs["device"]) speech = speech.to(device=kwargs["device"]) speech_lengths = speech_lengths.to(device=kwargs["device"]) # hotword self.hotword_list = self.generate_hotwords_list(kwargs.get("hotword", None), tokenizer=tokenizer, frontend=frontend)
