Mathematica 9 is now available
Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2008

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Possible bug in WAV export

  • To: mathgroup at smc.vnet.net
  • Subject: [mg87644] Re: Possible bug in WAV export
  • From: Szabolcs <szhorvat at gmail.com>
  • Date: Tue, 15 Apr 2008 05:47:58 -0400 (EDT)
  • References: <ftsd0o$ba7$1@smc.vnet.net> <ftv8on$7kv$1@smc.vnet.net>

On Apr 14, 11:39 am, Jerry <Jer75... at yahoo.com> wrote:
> The sound in Mathematica and the sound from the exported WAV file
> (using some unknown player) seem the same to me. I'm using
> V6.02 on a XP box.

I experimented some more and it turns out that the problem is not
limited to WAV (it is present with FLAC and AIFF too).  However, the
problem only appears randomly, and not in each evaluation. The
following command does not always give the same result:

ImportString[
 ExportString[
  Play[Sum[Sin[2 Pi 440 k t], {k, 1, 4}], {t, 0, 1},
   SampleRate -> 44100, SampleDepth -> 16], "WAV"], "WAV"]

In about 20 or so evaluations I always get a few incorrect results.
It is not even necessary to play the sound---it is possible to see the
difference by looking at the spectra:

Table[ImportString[
  ExportString[
   Play[Sum[Sin[2 Pi 440 k t], {k, 1, 4}], {t, 0, 1},
    SampleRate -> 44100, SampleDepth -> 16], "WAV"], "WAV"],
 {20}]

Can anyone else reproduce this problem?  It is very strange that the
result is not deterministic.  Any ideas why this might happen?

To demonstrate that this unbelievable thing really happens, here's an
another example, together with the result.  Note that the very same
command returns different results in subsequent evaluations:

In[1]:=
Table[
 Total[#[[1, 1, 1]]] &@
  ImportString[
   ExportString[
    Play[Sum[Sin[2 Pi 440 k t], {k, 1, 4}], {t, 0, 1},
     SampleRate -> 44100, SampleDepth -> 16], "WAV"], "WAV"],
 {20}]

Out[1]= {0., 268.745, -0.752563, -0.756226, -0.756226, -2.99866, 0., \
0., -2.98279, 0., 0.756226, 341.993, 0., -0.752563, 0., 268.017, \
-0.752563, 0., -2.2522, 0.}

The small fluctuations could be explained by dithering, but any
reasonably precise result cannot have a sum of 342!  The sound
function was constructed in such a way that the sum should be 0.


  • Prev by Date: Re: Numerical integration and list of points
  • Next by Date: Re: Deleting Integrate[] transformation rule (some progress)
  • Previous by thread: Re: Possible bug in WAV export
  • Next by thread: Re: Possible bug in WAV export