From time to time I have discussed the problem of synchronizing multiple audio devices. More and more often, these audio devices are becoming digital. The fact is, that there are fewer and fewer companies even manufacturing analog tape machines. Soon you won’t be able to purchase a new analog machine from anyone.

The two principal ingredients in synchronizing multiple devices is speed and time. There must be some way to tell where you are, and how fast you are going. With analog audio machines and video machines, the answer is easy. Video sync and SMPTE time code. Video sync provides a speed reference, and SMPTE provides time and speed reference information. With the advent of digital audio recorders and hard disk recording, two new parameters enter the game; word clock and trigger point.

SMPTE comes in several flavors; 30 fps, 29.97 fps, 25 fps, and 24 fps (fps= frames per second). In the visual world these refer to how many video or film frames are shown each second. Audio on analog tape does not occur in discrete chunks. The audio can be recorded at any speed from 1 7/8 inches per second up to 30 inches per second, and anywhere in between with the use of variable speed tape machines. If you only want to synchronize analog audio machines, you can use any rate of SMPTE that you desire. If you plan to use inexpensive synchronizers, the code must be the same on all machines, but more expensive synchronizers allow you to use different frame rates on different machines and still keep them locked together. You could have one machine running at 7 1/2 ips and another machine running with the VSO cranked wide open to 45 ips and they would stay synchronized just fine... Well, almost.

Any Week Now

Sometimes it is cumbersome to keep two machines running together throughout a project. Sometimes two machines won’t be available, and sometimes you just don’t want to wait for the machines to synchronize. Sometimes it takes 10 to 15 or more seconds for two analog machines to finally lock up. Most analog machines do not mute their audio while in the locking process, so you must endure the pitch changing and warbling of the slave machine until it is in sync. Sometimes this can be very distracting when you are trying to minimize run-up time while working on a guitar solo or vocal chorus. The answer? Slave tapes.

A slave tape is a tape that will be played on the secondary machine that chases the master. The slave tape can, however, be used on its own. If you intend on using a slave tape by itself, it must have enough information on it so you can hear all of the instrumental parts you need to do the overdubs. Since a lot of the instruments were recorded on the master tape, you need to make a rough mix of these tracks and copy them over to a minimum number of tracks on the slave machine. At some point the slave tape will no longer need to be used alone, and the reference tracks can be erased and the slave tape synced back up with the master.

If you synchronize the master and slave tapes before the reference tracks have been erased, you will hear phasing between the original tracks that exist on the master and the reference track of the same instrument that resides on the slave. This is because of the resolution of the synchronizer, the wow and flutter of the analog tape machines, and the accuracy of the SMPTE time code recorded on the tape. You can not get rid of the phasing completely, so everybody has secretly promised not to record multi-microphone images such as drums or stereo instruments with tracks that span machines. That way you will not generally notice the discrepancies.

OK, Can I Talk About Digital Now?

Here is where word clock comes in. Digital audio is recorded in chunks. CDs have 44,100 chunks per second. Digital audio on video tape has 48,000 chunks per second. When synchronizing two digital machines they must have the same number of chunks per second. (There are exceptions, but it gets tricky.)

Digital audio should be synchronized speed-wise by using the same word clock for both the master and slave. There is no phasing when you listen to a stereo track split between the two machines because all of the chunks line up exactly. All you have to do is get them to the same spot and they will stay together forever. SMPTE can be used for the purpose of getting to the right spot just like it is with analog audio.

Hard disk recorders keep tracks synchronized by keeping track of sample numbers. If the guitar part starts at sample number 4,800,000 and the drums start at sample number zero, then after the drums have played for 100 seconds, the guitar will play and line up exactly. When a hard disk recorder chases a SMPTE source, the sample number is mathematically calculated from the SMPTE location.. When the values match, the digital audio starts, thus the “trigger point.” There is a method by which digital hard disk sources can vari-speed, but we will skip that part for now.

It’s Getting Chunky In Here

With digital gear, all pieces in the puzzle must be locked to the same sample, or word clock, to operate without clicks and pops. In some instances I have seen two DAT machines act like they were synchronized, but there would be a slight click every few seconds. When the music was playing at a healthy level, it covered up the clicks, but when a piano chord was ringing out you could hear the clicks easily. The reason it worked as well as it did is because accidentally the crystals of both machines were close enough together to only drift off one sample every few seconds, at which time one machine would snap to the next sample chunk like a chain skipping a tooth on a clogged wheel.

When you make a digital copy from DAT to DAT, or ADAT to ADAT, it works because the recording machine is synchronized to the playback machine by a digital clock. In this case, the clock is hidden in the digital audio traveling from the play machine to the record machine. It would work just as well if word clock out of the play machine was connected to word clock in of the record machine. If you had a third DAT machine connected to the digital audio out of the second machine, the clock information from the first playback machine would pass through the number two machine on to the number three machine. In some studios I see multiple DAT copies made this way all of the time, but each machine in the chain increases the jitter in the signal and increases the possibility of errors being inserted in the chain.

The proper way to connect the machines would be to use a DAT machine with multiple outputs, a digital processing box with multiple digital outputs like the TC Finalizer, or some form of digital audio distribution device like a Z Systems 8x8 routing box, or the Ardvark Word Clock Distribution Amp. Now you have a single digital audio source from the playback machine that is cloned and distributed in parallel to as many DAT machines as you like. Same clock source, minimum jitter, the only way to go.

Remember to try to synchronize all of your digital machines, processors and digital consoles in parallel as close to the common source as possible, unless the source is video sync, then run one master device from video sync and feed word clock from the master device to all of the other digital audio devices. That way, only one device has to cope with the task of deriving a word clock signal from video.

There are some devices available that provide common clocks from a single source. Ardvark makes the Ardsync, Mark of the Unicorn makes the MIDI Time Piece AV and the Digital Time Piece, Digidesign makes the Universal Slave Driver, Opcode makes a new clock source called the Studio 64XTC, and N-Vision makes master clock.

Un-Syncable

Well, now that I’ve got you all worked up, I get that syncing feeling that I have to leave until next month. I have just touched the tip of the iceberg on a subject of Titanic proportions. And I am sure there are not enough life boats to save all of us. Me, women, and children first.