evaluation of speech enhancement algorithm. How?

> So you are trying to make the voice sound like you had not added the
> pink noise in the first place?

Yes.

>Can you give some idea what your
> relative speech/noise levels are?

SNR is 5dB

> Where did the whitening filter come from?

The whitening filter is necessary to obtain a signal that drives
the vocal tract filter.

> When you turn the noise level down and analyze/synthesize the waveforms
> do the original (clean) signal and reconstructed signal look similar on
> a point-by-point basis?

When there is no noise the original signal and reconstructed signal are 
equal to eachother.

Hi
Have you tried distance measures like the 'Itakuro-Saito distance
measure'?

Regards
Nithin

Lars Hansen wrote:
> Hello,
>
> What are the standard methods for testing the quality of the output from a
> speech enhancement algorithm?
>
> I read that segmental SNR can be used, but that requires that I choose a set
> of frames where I _know_ that the frame contains speech. Isn't that cheating
> anyway? I would argue that the variance of the difference between the clean
> speech signal and the estimated speech signal is a better performance factor
> as it truly tells you how well your estimated speech signal matches the
> clean speech signal..... or am I wrong?
>
> I am also interested in knowing if there are any programs out there that
> will process the output (wav-file) from my speech enhancement algorithm and
> give a score with respect to speech recognition. Is that possible?
> 
> Thanks in advance :o)

news:1128020548.868193.250250@g14g2000cwa.googlegroups.com...
> Hi
> Have you tried distance measures like the 'Itakuro-Saito distance
> measure'?
>


No....not yet...I have just downloaded some matlab-scripts to perform the 
measure you refer to. What is it about?

Hi
It is complex distance measure which compares two speech samples and
determines how closely they match. lower the measure , better the match
I guess. Although, while doing this it may be better to find the
distance measures of each of the segments and average them out using a
filter. In that case, you would know which segment actually behaved
very badly.

Nithin
Lars Hansen wrote:
> news:1128020548.868193.250250@g14g2000cwa.googlegroups.com...
> > Hi
> > Have you tried distance measures like the 'Itakuro-Saito distance
> > measure'?
> >
>
>
> No....not yet...I have just downloaded some matlab-scripts to perform the
> measure you refer to. What is it about?