[CLAM] about sms.m and CLAM FundFreqDetect

[Xavier Amatriain]

On Fri, 2006-07-21 at 15:43 +0800, Neediss Gu wrote:
>     BWT, SMS may can not handle the transient? Is it right? If someone
> has a binary of Sinsoid+Noise+Transient model, can I play it?

You can have a look at Tony Verma's publications
(http://www.acoustics.hut.fi/~tverma/publications/index.html ). It would
be great to add his extensions to CLAM. Also I know some people from
Philips had a different model that maybe even made it into MPEG4 but I
lost the reference (let me know if you find it).


> I have downloaded the sms.m and piloted it. I found that the first
> synthesis frame got some errors, the amplitude is so big, maybe
> somewhere has a bug. If you can find the fixed version, can you give
> me a copy? Thanks very much.

You can download it at
http://www.create.ucsb.edu/~xavier/temp/SMSMATLAB.zip


> ================ 2 CLAM FundFreqDetect.cxx ==============
> //------------------------------------
> "FundFreqDetect.cxx" line 175. 
> It may never be executed. And I can not catch the mean of x,y,a,b.
>  
> //------------------------------------
> "FundFreqDetect.cxx" line 339. // if(candidates2Error[j] >
> candidatesError[i]) 
> It may never be executed too.

If they are never executed it means that they are not very important :)
Now seriously I have cleaned up that file a bit but this new version has
still not been released. I will try to look into it if I have time but I
am afraid it is not going to be soon.

Xavier