MuseTalk
MuseTalk copied to clipboard
实时推理支持音频数据流
whisper_feature = audio_processor.audio2feat(stream) whisper_chunks = audio_processor.feature2chunks(feature_array=whisper_feature, fps=fps)
目前realtime_inference代码中的音频是基于音频文件来操作的,如何支持GPT-SOVits等实时音频流来进行实时推理呢?