Avoid leading/trailing silence in the RMS calculation

[achingacham]

The leading and trailing zeros (to represent silence) need to be avoided while calculating the RMS of the signal, as it impacts the overall signal distortion. This is particularly important when noise-mixing is performed with clean audio signals with different lengths of silence at the start/end of the actual utterance.

[Sato-Kunihiko]

Thanks.

I'll check it out later.

[Sato-Kunihiko]

@achingacham

This is particularly important when noise-mixing is performed with clean audio signals with different lengths of silence at the start/end of the actual utterance.

Can you give me an example of the audio signals? I'd like to know the specific example.