Unexpected background voice is detected.

[Alien1007]

I tried to process some audio clips collected from website with framewise detector following #12 . This model can detect human voice properly. However, I have found some results containing unexpected background voice tags, e.g. 'Background voice', 'Mechanisms', whose confidence is really high. I am sure there is no such voice in those clips. Do you have any idea?

[lmaxwell]

Unfortunately, I'm not sure what could be causing this issue. Could you please provide more details about the audio clips you're working with, such as their source and any specific characteristics that might be relevant?