DSPRelated.com
Forums

Some implementable VAD (Voice Activity Detection) algorithm or Single Channel Noise-Reduction Algorithm

Started by waheed khan March 27, 2007
Hi everybody,
I am working on designing a digital hearing aid as my MSc Project and am facing severe difficulties in removing noise from speech. There are loads of stuff about VAD on the web but all of them explain only the theory. I am at a very cruicual stage because time is running out. Can anybody help me in this regard. Plz its urgent.

Thanks in advance.
Hi,
If you are working with only speech. one of the easy and an effective VAD will be to auto-correlate your signal and look for atleast 3 consecutive equally spaced peaks to declare it as a "Voiced" frame.

Hari.

Hi everybody,
>I am working on designing a digital hearing aid as my MSc Project and am facing severe difficulties in removing noise from speech. There are loads of stuff about VAD on the web but all of them explain only the theory. I am at a very cruicual stage because time is running out. Can anybody help me in this regard. Plz its urgent.
>
>Thanks in advance.
If working on hearing aid, you might find this codec
interesting, since it's designed for such applications

http://focus.ti.com/docs/prod/folders/print/aic111.html
http://focus.ti.com/docs/toolsw/folders/print/aic111evm.html
http://www.digikey.com/scripts/DkSearch/dksus.dll?PName?Name)6-15335-ND&Site=US

That doesn't answer your question but the codec is
designed for hearing aids and is DSP comptible so
perhaps it could yield lower noise for you ...

waheed khan wrote:
>
> Hi everybody,
> I am working on designing a digital hearing aid as my MSc Project and
> am facing severe difficulties in removing noise from speech. There are
> loads of stuff about VAD on the web but all of them explain only the
> theory. I am at a very cruicual stage because time is running out. Can
> anybody help me in this regard. Plz its urgent.
>