I'm currently looking for an audio interface to buy. Yesterday I came across a YouTube channel that praised a particular maker of audio interface for "getting Linux" because they are up with the times and provide native packages including instructions for Omarchy.
On one hand the vendor giving first party support for Linux is great. But including Omarchy drops them down on my list.
@datenwolf I know this is not the point of this discussion, but shouldn't any decent audio interface just act as an USB Audio Class device? This stuff is standardized, there shouldn't be any need to install device-specific drivers.
@guenther as far as the actual audio signal goes, yes. But professional audio interfaces also provide sophisticated routing capabilities* which are not captured by USB audio class, but you got to configure somehow.
* for example: send and mix input 1 directly to output 4, which the headphone output 2 is monitoring, which goes to the headphones worn by the performer playing the instrument on input 1, so that they can hear their playing and the metronome coming from the DAW.
@datenwolf @guenther huh, and usb audio cannot represent this topology? perhaps it's less sophisticated than i remember it
Have a look at what alsa-scarlett-control implements. it's less about the USB driver itself (that's standard audio class) but the extended control commands the hardware affords.
https://github.com/geoffreybennett/alsa-scarlett-gui/blob/master/docs/iface-4th-gen-big.md
Depending on what you're looking for, it might be worth to have a look at devices that are categorized as digital mixers. Some have a USB port for audio I/O plus Ethernet for whatever protocol can be used to control them (OSC, proprietary shit). Just be careful not to get one where the USB audio stack is stereo-only.
@mildsunrise @datenwolf @guenther
You could (in theory) implement something like this in software and a bunch of cheap usb dongles... but you are going to be schlepping the audio data in from each device, into a software mixer, and then back out.
But, then if your computer lags or crashes, then that'll cause noticeable glitches in the sound.
And, in practice, I've found trying to delve into linux audio more frustrating and mystifying more often than not. (I've been meaning to try again at some point, especially with Pipewire and/or Jack)
Professional interfaces allow for a lot of common mixing tasks to be performed directly in the device itself, which is guaranteed to be low latency and resilient to software timing glitches.
@leon_p_smith @mildsunrise @guenther
Latency is the most important parameter.
Even under the best of conditions you're going to end up with latencies in the single digit millisecond range (number of samples in a buffer divided by sampling rate ×2 for the round trip + overhead). E.g. 96 kHz, 512 samples (32 bit, 2 ch makes a page) = 5.3 ms
Now that might not seem so bad, but adding as little as 5ms of latency in the monitor signal of a vocalist will "make them suck".
@leon_p_smith @mildsunrise @guenther
A note on sampling rate and bit depth: For the final mastered result 16 bits @ 48 kHz are perfectly fine, if the mastering engineer is competent.
Personally I like to have some extra 24 dB of wiggle room in the mastered audio, just so that minor screw ups in the mix don't turn into muddy results. That'd be 4 extra bits; and the next "nice" width is 24 bits.
For recording OTOH you want plenty of wiggle room in all directions. So 32 bits @ 96 kHz it is.