SMK1311

### Describe the bug from speechbrain.inference.speaker import SpeakerRecognition import torch device = "cuda" if torch.cuda.is_available() else "cpu" verification = SpeakerRecognition.from_hparams(source="speechbrain/spkrec-ecapa-voxceleb",run_opts={"device": device},) score, prediction = verification.verify_files("speechbrain/yinpin/原声.wav", "speechbrain/yinpin/line_1.mp3") # Different Speakers print('相似度得分:',score)...

如果将长视频分割成一分钟的短视频，短视频之间的大模型理解可以有上下文联系吗？