Bit-for-Bit Accuracy (Part 2 of 2)
July 17, 2007
The very useful process of frame synchronizing audio signals to re-time them to a local plant is good for PCM audio but will completely destroy other types of data such as compressed audio. Note that this applies to frame synchronizers that act on baseband digital audio signals as well as embedded synchronizers. This is due to the method employed by most all frame synchronizers to get the job done by re-sampling, also known as sample rate conversion.
Frame synchronization for audio basically involves accepting a 48 kHz signal and producing a 48 kHz signal that is synchronous with a local reference. The hidden part of this process is that it is not exactly 48 kHz in, or exactly 48 kHz out. Carried to several decimal points, it is more like 48,000.21 in to 48,000.03 out. It seems like a very tiny offset, but the two frequencies are asynchronous and equipment buffers will eventually have to drop or repeat samples to keep from over- or under-flowing.
This process works perfectly for PCM audio where the results of re-sampling are audibly insignificant. Compressed or data reduced audio uses the bits in an AES signal to carry data, and the resampling process basically scrambles the bits, thereby corrupting the signal. This prevents the audio from being decoded properly, or at all. If your facility is handling compressed audio, beware of sample rate converters--pals of PCM, killers of compressed.