By Todd G. Tatnall
DAW users often report this problem, "My song sounded fine when I tracked and mixed down. But when I burned an audio CD and played it back, the music was pitched down about a half-step and sounded slow. What happened?"
Let's start by looking at how this happens. We'll use Digidesign Pro Tools/Digi002 and the Presonus DigiMax for this example; however the problem can occur with just about any software and digital hardware if the conditions are correct.
The problem begins when a digital connection is made to the audio recording hardware. The DigiMax's eight mic preamps are sent digitally via ADAT lightpipe optical to the optical input on the Digi002. In order for the Digi002 to receive the digital audio signal properly, we need to set its "clock source" to "ADAT." This is necessary because each device has its own internal clock, which determines at what sample rate it should operate: 44.1, 48, 96kHz, and/or whatever. Just like two wristwatches almost never "tick" at the same instant, and often tick at slightly different speeds, digital audio sample clocks are seldom precisely the same. By setting the clock source of the Digi002 to ADAT, we are asking it to synchronize to the speed of the DigiMax internal clock through the optical ADAT cable. If we don't do this, we may experience noise, audio spikes, or lack of audio signal.
THIS IS IT
Because the Digi002 is now setting its clock speed by following the DigiMax, it's essential to make sure both items are set to the same sample rate. Here's where the problem occurs. Let's say the Digi002 and the Pro Tools session are set to operate at 44.1kHz while the DigiMax is operating at 48kHz. The Digi002 is synchronized to operate at the speed of the DigiMax clock; however, the Pro Tools software still thinks it's recording at 44.1k. Like most DAWs, Pro Tools records audio by creating individual audio files for each track. In addition to the audio data, the file contains information about how it was created, including the sample rate at which it was recorded. The result in this case is that files that were recorded at 48k are "labeled" as 44.1k files.
Anyone remember records? Vinyl? Near the end of their era, records operated at either 33 or 45 rpm. Suppose a record that was meant to operate at 45 rpm accidentally got a label that said it was 33 rpm. If you set your turntable to 33rpm and played the record, the music would be very slow and low-pitched. This is essentially what's happening in our sample rate mismatch scenario. The file was recorded at one sample rate, but improperly labeled with another.
Back in our digital recording world, the problem continues during mixdown. The engineer uses "Bounce to Disk" in Pro Tools to create a stereo mix of the audio tracks, keeping the Pro Tools sample rate at 44.1kHz for the bounce in order to comply with the Red Book audio CD standard. You can play this "mislabeled" stereo file back in Pro Tools (which is still using the Digimax 48k sample rate), and it sounds fine. (Remember that Pro Tools and the Digi002 think that they're operating at 44.1k while in reality the DigiMax is setting the speed of the Digi002 to 48k, making the file sound correct.)
But when the file has been burned to an audio CD, the slow, "pitched down" sound will be heard. When you play the disc back in a CD player, it's now playing at 44.1 kHz - slower than it was originally recorded.
WHAT TO DO, WHAT TO DO?
So are our tracks ruined permanently? Fear not, there are some things you can do. In our example, we can fix this right inside Pro Tools. Start by setting Pro Tools' clock source to "internal." Create a new 44.1k session and import the problem file. Select the Time Compression-Expansion AudioSuite plug-in, and set the Ratio field to 0.919:1 (the ratio of 44.1 to 48). By shortening the file, we make it play faster - as fast as it was originally recorded. Then using the AudioSuite Pitch plug-in, set the Ratio to 1.088:1. This raises the pitch of our selection by a half-step to its original state.
If you're not using Pro Tools, most DAWs/audio editors have similar tools for time compression/expansion and pitch shift. Unfortunately, these types of processes aren't always completely "clean" - you may hear some artifacts from the processing. But at least you're back at the right speed and pitch, and you've saved the files.
PREVENTION IS THE BEST CURE
Of course, the best way to fix sample rate mismatches is prevention. Always make sure that any devices connected to one another digitally (and the software that uses them) have the same sample rates. In our example, the Digi002 has sample rate lights on the hardware that indicate the current sample rate. If the Digi002 detects a large enough difference between the speed of its internal clock and the external clock (say 44.1 versus 48k), the sample rate light will blink. Watch for this sort of indication any time you set your clock source to external.
Todd G. Tatnall is the Senior Tech in Sweetwater's Technical Support department.
“Always make sure that any devices connected to one another digitally (and the software that uses them) have the same sample rates”