Is there a way I can record, hear myself, and not have latency with an audio interface?

[Majik]

This may be a good time to mention the challenges of using multiple audio interfaces. I hinted to this in my post above when I said

@Majik

There are good reasons for only supporting one device at a time (primarily around sample rate sync)

This issue here is that each audio interface uses it’s own internal timing. So even if you set two different audio interfaces to 48kHz (48,000 samples per second), the interfaces will be running at slightly more, or less than this. For instance, one interface may be running at 48,002 samples per second, and the other may be running at 47,997 samples per second.

Note that, by “audio interface” I am referring to any device that connects to the PC directly (e.g. via USB) for recording/capture purposes. So a Boss Katana or Fender Mustang, etc. recording via USB is an independent audio interface.

The problem that arises from this is that the DAW (and the PC audio system in general) needs to synchronise these two streams exactly. If it can’t then it will, every now and then, have to throw away samples from one or other of the audio interfaces.

The professional approach to this is to have audio interfaces “word clocked”, which means one AI acts as the master clock, and all other AIs use this as their clock. But consumer AIs don’t have this capability.

Historically, “pro-audio” drivers like ASIO and ALSA have avoided this issue by only allowing one audio interface at a time to be selected. And this approach tends to continue today.

However, there is another approach that has evolved over time, and it’s the approach used on the Apple Mac: adaptive resampling. Adaptive resampling is a way of resamplng one or more digital audio streams to a new rate, which then allows multiple streams to be synchronized.

Until fairly recently, adaptive resampling wasn’t very good, but with increased processing power and improved algorithms, it’s now much better. Some Apple Macs now have a hardware resampling built into their onboard audio interface (note this only works for connections to the inboard AI, such as SP/DIF inputs; it doesn’t work for external USB AIs).

Note that, whilst adaptive resampling is far better these days, it’s not perfect and it does introduce some minor distortion and additional latency (typically around 64 samples). However, as a cheap solution for multiple AIs, it’s far better than audio glitches. A professional recording studio would avoid using it, but for most home recording purposes, the impacts of using adaptive resampling for multiple AIs is usually negligible.

Apple Macs support the ability to configure using multiple AIs with adaptive resampling in a relatively user-friendly manner by creating an “aggregate device”.

This tends to not be so easy on Wndows or Linux.

ASIO on Windows doesn’t officially support multiple devices. There’s a number of approaches to bypassing that restriction, the best known of which is using ASIO4ALL.

ALSA on Linux does support multiple devices, but most audio apps will only connect to one ALSA device at a time. There are ways to enable multiple devices within Jackd, but it’s not particularly user friendly.

the good news on Linux is there is a new audio/video subsystem that has been developed called Pipewire which does support the connection and routing of multiple ALSA devices by default, and automatically handles resampling. Pipewire is becoming the default audio subsystem on many Linux systems, and is supported by GUI tools to make it more user friendly to use.

Interestingly, a major driver (pun intended) for Pipewire development has been Automotive Grade Linux.

Cheers,

Keith

[DavidP]

@Majik

@Majik

This could also go into a multi-input AI with direct monitoring. This should keep the tracks separate for recording purposes (if required) but allow a “monitor mix” in the AI going out to the direct monitor.

Thanks Keith. I’ve used the Direct Monitoring on my 2i2 when recording a guitar part over backing. But I’d never thought of using that when setting up to play and sing live using the Play Acoustic, I shall try that next time.

It may also work well when recording a vocal over backing to route the mic through the Play Acoustic and add just enough reverb and delay to produce a recorded track that already sounds a little more natural before adding fx in the mixing stage of production.

Couple of things to try when the opportunity presents.

Last post about adaptive resampling was fascinating.