[mirtoolbox] Re: Yet another newbie question

Hi Martin,

Your problem comes from a bug I just discovered in mirspectrum: when the 
computation is decomposed into chunks, the frames of the spectrum are 
reconcatenated in the end for the magnitude, but not for the phase. (The bug 
did not have any effect previously since it necessarily leads to an error 
message.)

This will be corrected in the next release. For urgent needs, here is the 
corrected evaleach.m file that can replace the one in @mirdesign folder.

Cheers,

Olivier



Le 15.8.2010 à 21.45, Martín Hartmann a écrit :

> Dear all, 
> 
> I am using MIRToolbox and Weka for some classification experiments and 
> therefore I am extracting some features and exporting the statistical 
> information to a Weka Attribute-Relation File Format (.arff). 
> 
> In previous tests I have been very satisfied using the Labeling option of 
> miraudio (page 19 of MIRToolbox manual) but in MIRToolbox 1.3 this doesn't 
> seem to work. But maybe I have written something wrong so I would be glad if 
> someone could provide me a working example of this option.
> 
> Further, I have noticed that the code in page 172 of the manual (Complex 
> Flowchart Design) does not work for me. For example when I write:
> 
>     myflow = mirstruct;
>     myflow.tmp.s = mirspectrum(‘Design’, ‘Frame’);
>     myflow.cent = mircentroid(myflow.tmp.s);
>     myflow.ceps = mircepstrum(myflow.tmp.s);
>     output = mireval(myflow, ‘myfile.au’);
> 
> This is the error I got:
> 
> ??? Error using ==> times
> Matrix dimensions must agree.
> 
> Error in ==> mircepstrum.mircepstrum>main at 101
>             m = mag{h}{k}.*exp(i*pha{h}{k});
> 
> Error in ==> mirfunction at 196
>     o = main(orig,during,after);
> 
> Error in ==> mircepstrum.mircepstrum at 44
> varargout = 
> mirfunction(@mircepstrum,orig,varargin,nargout,specif,@init,@main);
> 
> Error in ==> mirdesign.evaleach>evalbranch at 767
>     b = b.method(argin,b.option,b.postoption);
> 
> Error in ==> mirdesign.evaleach>evalbranches at 717
>     z.(fields{i}) = evalbranch(branch{i},tmp,y);
> 
> Error in ==> mirdesign.evaleach at 209
>             y = evalbranches(d,y);
> 
> Error in ==> mireval>evalaudiofile at 224
>     v = evaleach(d,single,name);
> 
> Error in ==> mireval at 137
>         yf = evalaudiofile(d,a{f},sr{f},w{f},{},0,f,single,'');
> 
> 
> I am using MIRToolbox 1.3 in Ubuntu Lucid Lynx and .au 22050 Hz soundfiles.
> 
> Congratulations for the MIRToolbox, I love it!
> 
> Thanks for your help, 
> 
> Martin
> 
>

Other related posts: