[linux-audio-user] Converting sample rate: failed...

[Erik de Castro Lopo]

On Wed, 22 Sep 2004 09:50:59 -0700
davidrclark at earthlink.net wrote:

> Thanks for your response.  I applaud your enthusiasm, but it appears 
> to me that you are laboring under some misconceptions 

That may be the case. I haven't seen your code so to a certain
extent I am comparing libsamplerate (which anyone can download and
evaluate) with something that has been seen and used by you alone.

> misunderstanding of what I have posted.  Among the misconceptions is a 
> serious one that should cause concern amongst those who use 
> libsamplerate.

Funnily enough the issue you see as a "serious misconception" I see
as a "significant advantage of libsamplerate over your converter". [0]

You have also said that libsamplerate is not good enough for you
to use on your own projects. If you want anyone to take that
remark seriously you should:

   a) Show us the code or provide a binary of your converter so 
      that your claims can be independently verified. [1]

and/or

   b) Provide some pathological signal for which libsamplerate
      fails to do an adequate job. [2]

No offense, but any response other than the above is just hot air.

Regards,
Erik

[0] You claim that the highly localized behavior of the truncated
    windowed sinc is a bad thing. I claim that this localization is
    a good thing for converting a general digital audio signal because
    these signals are already highly localized (ie snare drum hit at 
    10.03 seconds from the start). From your description, your 
    converter spreads these highly localized events over the whole 
    of the output signal which I think is a bad thing.

[1] I'd be intersted in testing things like SNR, bandwidth, pass
    band magnitude response, phase response, impulse response and 
    processing speed. I'd also like to do some listening tests.

[2] If you manage to find any pathological signal for which 
    libsamplerate does fail, I will fix the code and add tests for
    these cases to the libsamplerate test suite.

-- 
+-----------------------------------------------------------+
  Erik de Castro Lopo  nospam at mega-nerd.com (Yes it's valid)
+-----------------------------------------------------------+
"Lumping configuration data, security data, kernel tuning parameters,
etc. into one monstrous fragile binary data structure is really dumb."
- David F. Skoll