Hi August,
thanks for your quick reply and sorry for my slow one ;-) The short
answer is that cleaning up the headers of the soundfiles by
re-encoding with sndfile-convert did the trick. Celine, the student I
am working with, promised to write you a bit later with more details.
Best! Derek
On 6/15/10 5:31 PM, august wrote:
Derek,
Are you using MacOS X? What version of readanysf~? What version of gavl and gmerlin_avdec are packaged with it/used with it?
I don't suspect it is the soundfile itself, but just in case, can you put an example online for me. This is going to be a tough bug to find, unless you see some sort of regularity in how the sound turns to noise for you and can report that to me?
can you make a simple patch that isolates the bug?
Also, when the sound goes to noise, is it just that one particular soundfile that is used in readanysf or is it PD's entire output. In other words, when you hear noise, can you also hear the other readanysf's playing....or can you also make a simple osc~ and hear it play correctly?
The "Current file is either invalid or an unsupported codec." warning can also come if you send "play" to the readanysf object without it having a file loaded. I assume this is what is happening.
-a.
Hello August, list....
I'm helping a student's installation, and we have created a patch which uses 24 instances of readanysf~ to read from 24 different soundfiles between 15min and one hour in length. All sound files are mono, 16 bit, 44.1KHz WAV_PCM format.
The problem is that, after a length of time, the readanysf~ objects output noise rather than the soundfile. It is not the result of any single soundfile, and many or all of the readanysf~ objects can be affected by this simultaneously.
I have attached the abstraction in question. The object is instantiated as [readanysf~ 1], in other words the block size and and buffer size are defaults.
Sample terminal output is as follows while the patch is running:
Created new readanysf~ with 1 channels and internal buffer of 24 * 64 = 1536 Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec. Current file is either invalid or an unsupported codec.
Opening each soundfile individually with readanysf~ and sending "play" and "pause" messages reports no errors whatsoever, however.
I have sndfile-info data for all of the soundfiles. The only irregularity I see in this is one file which reports:
Unknown chunk marker at position 6087587. Resynching.
Besides that, most of the files report something like this:
File : F_DOK6.wav Length : 146725772 RIFF : 146725764 WAVE bext : 602 fmt : 16 Format : 0x1 => WAVE_FORMAT_PCM Channels : 1 Sample Rate : 44100 Block Align : 2 Bit Width : 16 Bytes/sec : 88200 *** minf : 16 (unknown marker) *** elm1 : 7506 (unknown marker) data : 140114520 *** regn : 92 (unknown marker) *** umid : 24 (unknown marker) *** DGDA : 6602919 (unknown marker) End
Sample Rate : 44100 Frames : 70057260 Channels : 1 Format : 0x00010002 Sections : 1 Seekable : TRUE Duration : 00:26:28.600 Signal Max : 11627 (-9.00 dB)
Someone suggested the noisy output may be the result of a buffer problem, but I am not sure who I could verify or correct this.
I checked with "top" while the noise was happening and saw no evidence that Pd was using any more memory than usual, and the CPU meter reported 30%.
Hardware is an Intel Mac Mini, software is Pd-Extended 0.41.4.
Any suggestions or other diagnostics I could run are appreciated.
Best! Derek -- ::: derek holzer ::: http://macumbista.net ::: ---Oblique Strategy # 136: "Remove specifics and convert to ambiguities"