More on Digital Desks

Responding to last week’s article about Digital Desks, Andrew Goodman-Jones
from Piranha Systems Pty Ltd has added to the discussion of Digital issues.

He writes:

There is a distinction between a desk with a “digital” user interface and one that does lots of digital signal processing.

There are three different concepts with the use of the word ‘compression’ :

* analog signal amplitude compression – processing to reduce the dynamic range of a signal – normally done in analog

* digital data compression in a lossless way – also called “data compression” in the IT industry.

* digital data compression in a lossy way – also called “perceptual coding”

A digital desk won’t mysteriously do the first one – amplitude compression, as this article seems to think.

Upsampling within integer multiples is fine – ie from 48kHz to 96kHz – not a problem in the world. Just every second sample is the same.

Increasing digital resolution within integer multiples is also fine – ie from 16 bit to 20 bit, 24 bit etc. Just bit shifting – no loss of quality.

If you’re downsampling or decreasing digital resolution within integer multiples, that’s fine but you’re just going to be limited by the lowest rate devices in the chain, but other than that, no loss in quality.

You can make as many conversions in the digital domain as you like without any loss of quality. (assuming integer multiple sample rates)

Also, using the second one, lossless digital compression, is fine. The same data that goes in comes out the other end – guaranteed. You won’t hear any difference in the sound because there is none.

THERE ARE problems when you convert from weird data rates like 44.1kHz… but what is anyone going to do about that? Stop using CDs? To convert it properly requires a lot of CPU time to re-map (or ‘dither’) the 44,100 samples onto a 48,000 sample space. Cheap systems will just drop samples that don’t fit or stretch samples to fill in the gaps – this will sound crap.

I guess the answer is don’t use 44.1kHz ever…. but with the crazyness of ANYONE useing lossy compression (MP3, ATRAC etc) in a broadcast chain … it doesn’t seem like they care anyway. These should only ever be used at the last link in the chain – ie the consumer’s player.

The article says: “He suggested that in some cases an analog desk might be as good or better than a digital desk because conversion and compression are only done once”

Well that’s assuming that the digital desk is converting back to analog to do the mixing and level control. I expect that if a desk is called “digital” that it would be full of Digital Signal Processing chips (DSPs) and that it would use proper mathematical methods of mixing – otherwise, I wouldn’t call it a digital desk – it would be an analog desk with DACs and ADCs on it.

So in summary, 44.1kHz and LOSSLESS audio compression (perceptual coding) are problems. Everything else about a digital desk is good if you’re working with digital signals.