OT: OS X Sample Rate Conversion Performance

[Cargoplex]

t my job, I deal with very large WAVE files (around 1.5gig) recorded at 24/96.

The client gets data CDs with the 24/96 files on them, and also audio CDs of each file. We use PCs in our sound lab, but I recently aquired two Macs from our graphics department. On the PCs, we use Wavelab's batch processor to convert the 24/96 files to 16/44.1. This usual averages about 3 minutes file.

So today, I set up a batch process in Peak to convert the 24/96 files to 16/44.1, and to my amazement/disappointment, it was taking up to 40 MINUTES a file! That is practically real time! The PCs we use are standard Pentium 4's with about a gig or ram. The Mac I use is a 733MHz G4 PowerMac with 1.25 gigs of ram.

I just don't know what else to say. In addition to this devastating revalation, I have also found Peak and the dicontinued Spark to be less than comparable to the power of Sound Forge or Wavelab.

I won't be giving up my Macs anytime soon, but if anyone has any insight on the these issues, please bring your two cents to this post. Thanks.

[bighairycellar]

It's probably more to do with Peak than OS X. But why would you want the conversion to happen quickly? Don't you care about the sound quality?

[Cargoplex]

Of course we care about the sound quality, but when you are dealing with 60+ files and the client wants the work done yesterday, you tend to count every minute. Note that once the CDs are created, there is also extensive databasing and labeling that must be done. I sure wish I could find some answers.

[Chris Cavell]

The time it takes really has no effect on the sound quality when you're comparing apples and oranges like this. The reason for Wavelab taking such less time is because the wavelab installation detects which brand and type of CPU is in the system and determines which code it should use to take advantage that specific CPU's instruction sets. (Waves plugs do the same thing on PC's, which is why they often have cpu-specific beta versions available on their update page almost always around the time a new PC cpu becomes available.) It's sort of like C compared to C++. You can have the same program written in either, but the code written in C will be much much much longer and take a heck of a lot longer to run b/c C++ includes commands/instructions that effectively replace tens, hundreds, sometimes thousands of lines of code that perform the exact same calculation. Many software developers do not try to get so specific to the cpu so as to ensure the broadest compatibility b/w computers, where others try to keep their stuff riding the front edge of technological advancements. It also has a great deal to do with the FSB speed b/w the mac and the pc's in question. They are both using 32 bit cpu's, but the minimum buss speed of the P4 CPU is 400MHz (the newer P4's go all the way up to 800MHz!), where that G4 is 133MHz. Some of it also depends on the speed of the RAM, chances are that the ram in the pc's is operating at 266MHz or higher, where the G4's ram is operating no higher than 133. Listen to the quality of a short snippet done on both machines and you'll probably hear no difference in quality at all. I'm not saying PC's are better than Mac's or vice versa, just merely pointing out the specs of the systems you're using and how they affect the speed of processing audio.