[Clam-devel] Pitch transposition network and a new network for pitch discretization

Pau Arumi parumi at iua.upf.edu
Fri Jun 22 06:00:33 PDT 2007


Xavier Amatriain wrote:
> Hi all, I am still in the middle of moving to Barcelona and still don't 
> have any working CLAM Sandbox.
> 
> I am also surprised that it crashes on that particular line in 
> SynthSineSpectrum. However, this file changed
> recently (I moved precisely the BlackmanHarris transform table to 
> another file). 

wasn't it these commits?
r10037 | xamat | 2007-04-21
Removed huge window table from this file.
------------------------------------------------------------------------
r10036 | xamat | 2007-04-21
Removed huge window table from the SynthSineSpectrum.cxx file and 
moved
it to a separate file.

this is much older than 1.1 (r10216) which seems to be ok.
will keep searching...

pau


All I can say is that some debugging
> is needed! Is the exception an out-of-bounds read in the array? How is 
> the heap and local variables?
> 
> 
> 
> Pau Arumi wrote:
>> En/na Hernán Ordiales ha escrit:hi all,
>>>
>>> I'm working in the pitch discretization network taking pitch
>>> transposition one as example. Aside still no working yet (starts and
>>> stop suddenly, I've to check pitch shift), a new problem arise, in the
>>> last svn version, at least for me, the pitch transposition network
>>> works for a while and then crashes and has only noise at output (and
>>> displays lines in the oscilloscope, like plotting complex values, no
>>> peak visualization any more)
>>
>> The same here!
>> I've started a new network from scratch that uses SMSTranspose.
>> It seems to run fine when no control is sent. Once you start sending 
>> control changes the audio doesn't change (transposition does not work) 
>> and when control value gets around 1000 it crashes 
>> CLAM::SynthSineSpectrum::FillSynthSineSpectrum
>>
>> The backtrace:
>> (gdb) bt
>> #0  0xb7ad7f8d in CLAM::SynthSineSpectrum::FillSynthSineSpectrum 
>> (this=0x8733c80, peakArray=@0x875ea90, gain=1)
>>     at scons/libs/processing/src/SynthSineSpectrum.cxx:164
>> #1  0xb7ad85fa in CLAM::SynthSineSpectrum::Do (this=0x8733c80, 
>> peakArray=@0x875ea90, residualSpectrumOut=@0x86e3258, gain=1)
>>     at scons/libs/processing/src/SynthSineSpectrum.cxx:89
>> #2  0xb7ad9be1 in CLAM::SMSSynthesis::SinusoidalSynthesis 
>> (this=0x8731df8, in=@0x875ea90, outSpec=@0x86e3258, outAudio=@0x8734974)
>>     at scons/libs/processing/src/SMSSynthesis.cxx:168
>> #3  0xb7ad9d45 in CLAM::SMSSynthesis::Do (this=0x8731df8, 
>> inputSinusoidalPeaks=@0x875ea90, inputResidualSpectrum=@0x8705ee8,
>>     outputSinusoidalSpectrum=@0x86e3258, outputSpectrum=@0x87408f8, 
>> outputAudio=@0x8734804, outputSinusoidalAudio=@0x8734974,
>>     outputResidualAudio=@0x87348bc) at 
>> scons/libs/processing/src/SMSSynthesis.cxx:251
>> #4  0xb7ada20b in CLAM::SMSSynthesis::Do (this=0x8731df8) at 
>> scons/libs/processing/src/SMSSynthesis.cxx:182
>> #5  0xb7f4f5bc in CLAM::NaiveFlowControl::Do (this=0x86bce90) at 
>> scons/libs/core/src/NaiveFlowControl.cxx:122
>> #6  0xb7f47d70 in CLAM::Network::Do (this=0xbfb8aa34) at 
>> scons/libs/core/src/Network.cxx:440
>> #7  0xb7e649a8 in CLAM::JACKNetworkPlayer::Do (this=0x86c0970, 
>> nframes=1024) at scons/libs/audioio/src/JACKNetworkPlayer.cxx:244
>> #8  0xb7e67e90 in CLAM::JackProcessingCallback (nframes=1024, 
>> arg=0x86c0970) at scons/libs/audioio/src/JACKNetworkPlayer.cxx:14
>>
>>
>> But there are no significant changes in SMSTranspose
>> from 1.1 release, only the factory registration changed.
>>
>> On the other side the sintracks plot out of smsanalysis already
>> shows this strange single-peak. The problem seems in smsanalysis,
>> however SMSAnalysisSynthesis back2back tests are green. I'm
>> puzzled...
>> Ideas, anyone?
>>
>> pau
>>
>>
>>
>> _______________________________________________
>> Clam-devel mailing list
>> Clam-devel at llistes.projectes.lafarga.org
>> https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel
> 
> 
> 
> _______________________________________________
> Clam-devel mailing list
> Clam-devel at llistes.projectes.lafarga.org
> https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel





More information about the clam-devel mailing list