Telephone colophon: Or, how I overengineered my call audio (2020)

I really wish it was easier to do high-quality audio calls these days.

It's technically feasible, apps that can do this have existed for years[1,2,3], but they're either non-free or kludgy and unintuitive as hell.

At this point, It's definitely a UX problem, not a "we don't have the tech to do this" problem.

Analog phones in the 80s sounded better than almost anything a typical consumer is likely to interact with these days[4]. Now, it's all crappy 16kHZ Bluetooth headsets, bad noise / echo cancellation everywhere, and all that encoded with some low-quality opus.

nobody seems to care about this very much. We now have devices that can go up to a few hundred mb/s over WiFi, yet Bluetooth hasn't changed much since 20 years ago, and the audio quality is basically what it was back then.

[1] https://bearware.dk [2] https://support.zoom.com/hc/en/article?id=zm_kb&sysparm_arti... [3] https://cleanfeed.net/ [4] https://evan-doorbell.com/wp-content/uploads/Overview-rough....

That's a lot of software setup for something that can easily be done in hardware. I've been over engineering things for years now.

For my mic I have also an sm57 feeding into a dbx286 hardware mic preamp and dynamics. Then feeding that into my audio interface for calls that also gives me a knob to mix in my mic into my headphone signal. This all gives me zero latency monitoring off all the gates and compression. Then for the output signal I have a seperate audio interface I use with all the web calling applications that sends the audio of all other people in the call to o another compressor that levels out their volumes. Then that is sent through a tc electronic multiband compressor unit to fix the really dull mics some people have. This is then mixed with the stereo output of all other applications on my audio interface again.

This way I have consistent audio no matter what OS I'm booted into at that moment.

I've tried several different mics but eventually settled on a wired headset and Revelator io44 audio interface. The latter one is a goofy brick but it has a TRRS audio jack and built-in DSP so I don't have to fiddle with loopbacks, DAWs and VSTs.

And if I'm not able to lug that brick I can just plug the headset directly into my laptop.

This is basically what services like cleanfeed[1] are designed to do: send and receive high fidelity audio with minimal latency.

I'm sure in practice you can't have a jam session more than 50ms or so away from your bandmates before the latency starts becoming very noticeable.

[1] https://cleanfeed.net/

Reminds of of working with ISDN lines to board eons ago ;)

I guess that's one way to do it. For a simpler approach, I'd just use a basic 4-chan mixer and feed its output into the computer.

Reading this it turns out I have overengineered my call audio for years without even being aware of it.