-
|
May I ask if this vad supports detecting only nearby voices while ignoring the conversation of people around? |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
|
it is not designed for far-field speech per se |
Beta Was this translation helpful? Give feedback.
-
|
one criterion may be - intelligibility of speech |
Beta Was this translation helpful? Give feedback.
-
|
Your answer is great. Real-time call products on the market, such as ChatGPT and Doubao, seem to have near-field VAD detection capability in my tests, which is not achieved through voiceprint detection. Perhaps, as you said, they rely on theintelligibility of speech |
Beta Was this translation helpful? Give feedback.
it is not designed for far-field speech per se
emphasis in the latest release was to suppress background speech without making the overall performance worse
the problem is that there is no real criterion to distinguish foreground and background speech
loudness is not an option because audio can be non-normalized or just quiet as an input
and in a streaming scenario it is difficult to normalize