ten-vad seems unable to cut in faster speech scenarios

[twmht]

I actually compared ten-vad and silero-vad under conditions of faster speech to see which one could better cut at sentence pause points. I set ten-vad to cut whenever the pause exceeded 20ms, while sirelo-vad was set to cut at 40ms. The result was that for the same speaker speaking continuously for thirty seconds, ten-vad did not cut, whereas sirelo-vad performed better, cutting several sentences.

My test language is Chinese; is ten-vad's support for Chinese not as good?

[Ziyi6]

Hi, in our large testset, our ten-vad performs well. For better analysis of the result, would you mind to share some of your test wavs, and if possible, could you share your cutting script as well?