🚀 go-pugleaf

orion@proxima.cc.colorado.edu  writes:

>  I am working on a Juke Box program and am having a couple of problems.
> 
>  First, I do not always receive the didPlay: message that should be sent by
> a sound when it stops playing.  This causes my program to stop, and not play
> the next song.  This behavior is erratic, i.e. sometimes I get the message,
> sometimes I don't.  The more I am doing elsewhere (the more in the background
> the program is) the less likely it is to get the message).
> 
>  Second, I am getting a lot of drop outs when I play the songs.  This may be
> 
> unavoidable as the songs are stored in compressed format on a NFS mounted
> drive, but any suggestions would be appreciated.
> 

I've had similar problems to yours.  My repeated sounds were receiving TIMEOUT  
errors after a few repeats.  The problem for me seem to occur only with sounds  
I recorded using the microphone.  I discovered a solution by looking at the 3.0  
release notes (/NextLibrary/Documentation/NextDev/ReleaseNotes/Sound.rtf) where  
it says:

   "Compressed soundfiles must have either 22 kHz or 44 kHz sampling rates in  
order to be playable without first decompressing.  In particular, an 8 kHz  
CODEC file (such as from the built-in microphone) must be first resampled to 22  
kHz (e.g., by SNDConvertSound()) before compressing.   Since ATC format  
discards empty portions of the sound spectrum, upsampling does not increase the  
file size as you might expect."

I have fixed the problem by converting those sounds (at runtime) to a different  
format before playing them:

	if ([mySound samplingRate] < SND_RATE_LOW)
	{
		int err;
		
		err = [mySound convertToFormat 
			:SND_FORMAT_LINEAR_16
			 samplingRate:SND_RATE_LOW 
			channelCount:2];
#ifdef DEBUG
		fprintf(stderr,"Converted? %s\n",SNDSoundError(err));
#endif

	}

Performance still does not seem as good as with sounds that don't need to be  
converted.  I wonder whether some shenanegans are occuring where the main  
processor is being used instead of the DSP.

But this does seems to fix the main problem.


-- 
Doug Kent
slugg@mouthers.wa.com