SMPTE Part 1

Christmas 2008 was my first experience diving into SMPTE, MTC, and VTR decks so that we could sync our primary Beta deck with ProTools for audio track playback.  In order to start this discussion, I thought it would be a good idea to start out with SMPTE 101.  Brent Hoover wrote a great overview over at the study hall on ProSoundWeb.  Below is a summary of his thoughts…

What is Timecode? 
Timecode is simply a way of recording a “time” along with either a visual or audio event. This allows you to say, “The beginning of the song is at 0 and the cymbal crash happened at 2 minutes and 53 seconds.” The information is then stored along with the audio or video medium so that, as it is moved from device to device, the time information stays the same.

The most commonly used timecode format was developed by the Society of Motion Picture and Television Engineers. Referred to as SMPTE (pronounced Simp-Tee), it was originally developed to synchronize sound for film but has since been adapted for video as well as audio-only use.

The other commonly used timecode format is Midi Time Code (MTC), which was developed by Chris Meyer and Evan Brooks for Digidesign.

How does Synchronization work? 
Imagine that you and a friend have the only two watches in the world, so that you have no time reference other than your own watches. You need to meet at an exact time later that day. No watch is exactly accurate – maybe yours is a bit slow and your friend’s is a bit fast. Over the day the result is that there is a 10-second difference between your two watches. While it may be okay with your friend that you’re 10 seconds late because he likes you, it would be very bad for a cymbal crash. In audio, even delays of a fraction of a second with doubled vocals can be noticeable.

The next day, in order to circumvent the 10-second time difference, you and your friend work out a system. You pick one person as the Master and one person as the Slave (it’s that kind of friendship). The Master calls the Slave on his cell phone and tells him what time it is every few seconds, and the Slave resets his watch to match the Master’s. While this system may be an annoying prospect for human beings, it works quite well for machines. Similarly, in every timecode set-up you must determine which machine is the master (there can be only one) and which are the slaves. Usually the constraints of the machine will define who is master and who is slave. 

How would I use Timecode? 
Sync-to-Video - The simplest video sync scenario works much like this: a VHS tape has LTC on one of its audio tracks (it is the Master). The device to be synced (the Slave) reads that timecode and chases it. The tricky part is determining which type of SMPTE frame rate to use.

When using SMPTE at 30 frames-per-second (fps), one second equals one second. Unfortunately, NTSC color video does not run at 30fps — it runs at 29.97fps. Why in blazes does it do that, you ask? Well, let’s just say that video, like life, was simpler when it was in black-and-white. But when color was introduced, they decided to cram that little bit of color information in the space left by .03 frames.

At 29.97fps, a second of material is no longer a second, but just a hair longer. While a few hairs may not matter to you and me, they matter a lot to people in television who make their living selling seconds, so these people came up with a solution called “drop-frame”. This means a few frames are dropped so that SMPTE time matches with real time.

So we have SMPTE at rates of 30fps, 29.97 “drop” and 29.97 “non-drop”. All that really matters on a basic level is that you need a tool that can handle all these different frame rates, and you must know exactly what frame rate the video you are syncing with is using.

In the next post on SMPTE, I’ll go over our setup and implementation, along with some learnings from an intense production process.