Wireless lossless digital audio

I was in the cable industry for over 20 years and am well acquainted with NTSC ... I suspect the audio quality at 6 MHz to be far better than an FM signal which is nominally 100KHz wide. The FM system is more resistant to amplitude changes (due to distance, etc) but can't encode as much data as the VSB-AM signal used by NTSC.

There are a few caveats and I haven't looked at the transmitter/receiver pair he purchased. If it's simply 6MHz of bandwidth transmitted for some specified distance, it will probably work. SPDIF is a digital signal and while it runs at 6Mbps, there are significant harmonics that can be passed through the fiber. Since we can assume the first harmonic will pass through the link, you can expect a signal which has more sinusoid pulses coming back out of the receiver. With a PLL to recover the clock and a good data recovery circuit added to the receiver, I'm guessing he could easily push this link to 50 to 100 feet.

As an aside, the NTSC signal is actually sending far more information as a whole and while it uses an AGC to overcome amplitude changes during transmission (the sync pulse is adjusted to be -0.3V), it also uses a comb filter to recover chroma information and the luminense is AM modulated above at the pixel clock rate. Your eyes are actually pretty low resolution compared to your ears and there are lots of ways to fool the brain's image processing (e.g. interlace makes a 30FPS system look more like 60FPS).

In any case, a very creative build!

16 bit is not just adequate it's optimal. It's enough for perfect sound, unless, as you point out, you are editing.

Disappointing this pseudo science makes it on to HN.

There was a nice article here a while back that went through the entire subject in depth, can't recall the site.

Pro audio uses high bit depth throughout the signal chain to simplify gain staging. 16 bit is just barely adequate for normalised signals, but it is inadequate for signals of unpredictable level as it leaves no room for error. Set your gain too high and you get clipping; Set it too low and you raise the noise floor. You never need 120+dB of dynamic range, but it is incredibly useful to have 30dB of headroom to cope with sudden increases in level and an inaudible noise floor.

This extra headroom is critical for digital systems, as they have no headroom above 0dBFS - most analogue devices distort progressively (and often rather pleasantly) when pushed too hard, but digital clipping sounds horrible. Once you've run out of bits, all hell breaks loose.

16 bit is absolutely fine for domestic audio, but it is totally inadequate in a production environment. Without the extra dynamic range provided by 24 bit, you're stuck with some kind of compromise - either you risk overs, you raise your noise floor, or you have to stick limiters in front of all your ADCs and put up with your peaks getting squashed.

Yes, Apple really needs a low latency protocol for both audio and also AirPlay video. Would essentially enable any device to be a game console.