Re: [Audacity-devel] windows, noise and dtmf tones

[<Mar...@ao...>]

 
Hi Salvo
 
Didn't get your first email, only the second.  Took some time to  disentangle 
the patch from the email mangling it got but I have applied it and  run it 
and overall I'd say it's a pretty good effort!  Didn't do the second  one, 
however.
 
A few comments to (hopefully) help you on:
 
If you model NoiseDialog::MakeNoiseDialog on  
ToneGenDialog::PopulateOrExchangeStandard in ToneGen.cpp I think you'll find it  much simpler, and won't have 
the vertical layout problem that you have at the  moment (labels don't line 
up with boxes).
 
EffectToneGen::EffectToneGen() shows how to set up the parameter defaults  
better, so that they get remembered each time you use the noise generator.
 
The choice of the pink noise filter is, I guess, OK (I don't know much  about 
this, other than what I just read) but with the default amplitude of  0.8 it 
leads to clipping, which must distort it's spectrum.  This can't be  good.  
Why not turn it down by about a factor of about 4?
 
In your comments for the brown noise generator, you seem to assume that  
fs=44100; this is not necessarily the case, it depends on the project  rate.  
Should fc very with fs?  I don't know, but suspect it  should.
 
I have not checked the new window functions in detail but they basically  
seem to work.  Wish I'd got around to fixing the vertical scale problem in  
Frequency Analysis!
 
Attaching patches to emails does not work very well.  Can you post  them on a 
web server somewhere for download?  Also, I don't know if it's  just me but I 
have to run 'patch' in the appropriate directory and so had to  split the 
patch into two, one for each directory src/ and src/effects/.   Since I had them 
done, I posted them (temporarily) here _http://members.aol.com/martynshaw/_ 
(http://members.aol.com/martynshaw/) 
 
All for now, HTH
Martyn
 
In a message dated 08/11/2006 07:32:25 GMT Standard Time,  
ven...@ya... writes:

Hello  again!
and here a pre-version for the Noise Removal, which now
allows to  use different window functions as well. 

This is more "experimental",  in the sense that I would like
some feedback as for the "alfa" thing -  there is some
relation between higher alfa, lower window lenght,  sampling
frequency - and as for the usage. Also, I did not take care
of  the ClearSpeach case - what to do there? ...

thx!
.salvo

ps.  in order to apply this patch, you will need to apply
also the ones of my  other email. I believe only those for
the FFT are needed, but haven't  double checked.





--- salvatore ventura  <ven...@ya...> ha
scritto: 

>  Hi,
>  so here we go: different files in here, they do the
>  following:
>  
>   - the "generate noise" with more  options
>   - additional window functions available in the  FFT
>   - new window functions available for use in the  Analyze
> 
>     Spectrum
>   
>   Hope the format is ok. Files are DOS mode: if needed,  I
>  can generate UNIX ones.
>  
>   Let  me know,
>  
>   thx,
>    .salvo
> 
> 
...