[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: VAD (Voice Activity Detection) algorithms?



VAD is perhaps not a bad place to start ... how does one know after all
whether an auditory object exists or not, other then some sort of
activity detection.
I can suggest a VAD which is part of a signal de-noiser. Assuming you
have one stream and know that the initial N frames of the stream have no
auditory object embedded (only noise), then you can de-noise the signal
and asses VAD all at once ! Check here :
http://www-sipl.technion.ac.il/flatmax/speech/index.html

On the topic of individual signal separation, as said by others, there
are a huge amount of methods which are used ... none of them really
compare to mammalian separation quality currently, this is a part
quantitative and part qualitative judgment. None the less, radar systems
are capable of some degree of physical source identification and then
separation. I would guess that some of the methods of radar signal
processing would lend methods to CASA. Strange as it may seem !
Further some people believe that closer modeling of the processes of
hearing lend to reveal the methods developed through evolution of sound
source separation and identification.

Matt

On Fri, Apr 30, 2004 at 07:26:26PM +0100, Richard H. wrote:
> Good idea!
>
> I'd forgotten about all the goodies - including source code - in the GSM etc standards.
>
> Thanks,
>
> Richard
>
>
>   ----- Original Message -----
>   From: chen zhixin
>   To: AUDITORY@LISTS.MCGILL.CA
>   Sent: Friday, April 30, 2004 6:55 PM
>   Subject: Re: VAD (Voice Activity Detection) algorithms?
>
>
>   Hi, Richard
>
>   Both ITU G.723.1 and G.729 provide VAD algorithm/c code. They perform well in modest SNR environment.
>
>   Best Regards,
>   Chen
>
>   "Richard H." <auditory@AUGMENTICS.COM> wrote:
>     Hi,
>
>     Does anyone have any idea where I can find some simple algorithms/code to allow the presence or absence of speech in a signal to be
>     detected?
>
>     Thanks,
>
>     Richard
>
>
>
>
> ------------------------------------------------------------------------------
>   Do You Yahoo!?
>   ?Y??TT????????????????????????D??????????
--
http://flatmax.org

WSOLA TimeScale Audio Mod  : http://mffmtimescale.sourceforge.net/
FFTw C++                   : http://mffmfftwrapper.sourceforge.net/
Vector Bass                : http://mffmvectorbass.sourceforge.net/
Multimedia Time Code       : http://mffmtimecode.sourceforge.net/