01_audio_chunking.py 在处理多通道的音频的时候能量计算有误

[Le-Xiaohuai-speech]

01_audio_chunking.py 第 40 行 energy = np.mean(np.sum(segment**2, axis=1)) / len(segment) 这个算出来能量太小了

[chengjt23]

Hi @Le-Xiaohuai-speech,

Thank you very much for pointing this out!

The root cause was that our original pipeline was designed with a pre-processing step that ensured all audio was mono, so we didn't fully account for multi-channel inputs in this script at the time.

We have now updated the code to use energy = np.mean(segment**2) in commit 564d994, which ensures the energy calculation is accurate for both mono and multi-channel audio.

Thanks again for your valuable feedback!