OBS + Facebook Live workflow

[AaronD]

once you get this setup down pat, you may want to start optimizing audio (compression, etc), and that will take CPU (you have plenty of CPU, no issue there), but still...

Minor quibble there. Audio barely takes anything, even on the machines that I grew up with. 300MHz AMD K6 for one of them, that did a pretty good job with Windows 98. :-)

Now, if you're coming from the world of microcontrollers, where the typical machine is 8 bits wide, runs at 20MHz, and has maybe 1kB of RAM total, yeah, you'll need something better than *that* to do audio! But for a 32-bit or 64-bit machine, even with only MB of RAM and a few hundred MHz (fractional GHz) of clock, audio is almost nothing. At least, until you start mixing Bohemian Rhapsody on it or something like that (100's of tracks), by which point you'll have abandoned OBS a long time ago!

---

I also had a DAW on an old laptop. 700MHz Pentium 3. That DAW itself was hard-coded for 16 tracks, but it did just fine on that machine. I used it to practice mixing music: render some classic rock MIDI files into a bunch of WAV's, load those WAV's into the DAW, and have some fun. :-)

It helped a lot to render the guitars using the "Clean Guitar" MIDI patch, and then add the "fuzz" back in as an audio effect. MUCH better than the "Distorted Guitar" and "Overdrive Guitar" MIDI patches! And it actually did some justice to Deep Purple's fuzzy Hammond as well! (instead of the MIDI file's attempt to double an organ patch and a distorted guitar patch playing the same part)
And that old laptop would do that too, in addition to playing 16 WAV's simultaneously and mixing them. And I put some light reverb on the master too. Still did just fine.

The amount of processing power that that laptop had, is probably in the single digits of % on a modern machine. And it did all of that smoothly.

 

[Lawrence_SoCal]

Completely agree that audio processing is orders of magnitude less computationally demanding than video.
I was coming from some recent audio plugins/filters/effects that I presume are inefficiently coded as they are CPU hogs... shouldn't be...but are. On the other hand... noise cancelling is modern, and much more demanding than simple compressions, etc. just something to be aware of
Aaron - you'd argue a better Mic and speaker setup to begin with, and you'd be right. Done right, audio processing is minimal CPU impact ... but for those looking for brute force, not-that-great-sounding compromise of ease of setup without changing room, mics, speakers, etc work-with-you-find... some CPU intensive audio processing may be appropriate

 

[AaronD]

Completely agree that audio processing is orders of magnitude less computationally demanding than video.
I was coming from some recent audio plugins/filters/effects that I presume are inefficiently coded as they are CPU hogs... shouldn't be...but are. On the other hand... noise cancelling is modern, and much more demanding than simple compressions, etc. just something to be aware of
Aaron - you'd argue a better Mic and speaker setup to begin with, and you'd be right. Done right, audio processing is minimal CPU impact ... but for those looking for brute force, not-that-great-sounding compromise of ease of setup without changing room, mics, speakers, etc work-with-you-find... some CPU intensive audio processing may be appropriate

Yeah, some of the modern processing takes more CPU than just the "analog emulation" sort of stuff does. Echo cancellation, noise suppression, etc. But even that disappears compared to video.

And yes, I would argue a better mic and speaker setup, but it doesn't have to be expensive. In fact, it can even use what you have already. Just know how it works - look up their 3-dimensional sensitivity patterns, imagine all of those in the room together, folded at "mirror surfaces" (anything hard), and don't have them overlap, for example - and you'll do lightyears better already than if you just toss them up and wonder why they insist on misbehaving. Same for tracing a given signal through all the processing to see where it comes out, and noticing a feedback loop in the process. Or tracing backwards to find everything that feeds a given destination, and killing an unused source of noise in the process.

AI has some great promises, but at the moment at least, the privacy concerns are going to dominate. Sort out the problem of massive intrusion for massive profit, then let the panic settle down from that, and THEN we might start to make some decent advances in things that are actually useful, like capturing a perfectly clean whisper across the stadium during a rock concert. (slight exaggeration of what people tend to expect from their conference calls, but you get the idea)