Hacker News Reader

Comment by hackingonempty | Hacker News Reader

For what application do you need 96kHz or 192kHz of bandwidth?

After ridiculing 96 kHz sampling for years, I finally found a use for it.

A niche use, mind you.

Basically, I found myself doing real-to-complex baseband conversion for audio signals, and doing so efficiently halved my Nyquist rate.

Increasing my sample rate to 96 kHz let me construct the 48 kHz analytic signal that I wanted.

Hoisted on my own petard :)

Ultrasonic recordings

https://www.youtube.com/watch?v=hCQCP-5g5bo

Don't forget about cats and bats, they like music too

Ultrasound. I need to record and process signals up to 80kHz, so I need 192kHz sampling. No way around it.

With stage audio equipment?

Good old FM radio (in stereo) is a 192kHz mux, annoyingly.

How so?

Stereo pair, pilot, RDS (the scrolling artist/song text), and the 67 kHz SCA. It's not quite 192kHz as I recall, but that's nearest fit really. AES192/192kHz MPX composite on the back end of every exciter/transmitter when I was last near them.

That's for the FM baseband signal, not the audio. Among other things you're doing here, you're basically treating a stereo signal (plus a pilot that effectively contains no information, plus an extremely low bitrate RDS stream in an extremely inefficient way) as one monaural signal.

But maybe there's a use case for replacing an AES192 signal carrying the full FM baseband signal over ETH-67? I'm not sure it's the correct fit, though.

That's how radio works now, it's owned by like 3 companies. Automation/playout for large groups of stations is all centralized, and they carry the baseband over IP to transmitter. Radio solved it awhile ago.

https://www.telosalliance.com/radio-processing/audio-interfa... etc.

That's may be how the back-end of a modern broadcast FM radio transmitter works, but the good old stereo radio transmission itself still the same as it's been for many decades: It is analog, and mid-side encoded, and it remains completely compatible with monophonic receiver implementations.

Tuning into time radio signals like DCF77?

Recording music?

Distributing music != recording music != processing music.

There ARE good reasons for recording and processing music at higher bitrates and sample sizes. Effects, especially those with positive feedback loops, can go more unstable and clip and lose information with fewer bits and samples.

There's just no good reason to DISTRIBUTE music at the higher rates.

Don't such effects already upsample and downsample as needed internally ? You don't need to waste cpu on the effects that don't need higher sampling rate, right ?

They could. But then you get clipping effects and sampling effects when you convert back and forth.

Better to upsample once (generally a non-lossy operation) and operate on single precision FP (24-bit mantissa and an exponent) at higher sampling from that point forward and then downsample once (generally a LOSSY operation).

Reply on news.ycombinator.com