Ableton Forum

Hi all,

How do you guys deal with audio "pops" when loading in instruments or changing audio output of a track e.g. ?
Even when you enable a device pops happen, for some reason live11 and 12 have mayor cpu peaks for some reason..
So having multiple instruments on a chain and turning devices on when needed will create a click (cpu spike).

I'm using a buffer size of 256 samples but I feel it's pretty unstable in a live set.
Any higher makes live performing (guitar and vocals) impossible due to latency.
The limitations are driving me mad.

---

I could of course go for a higher buffer size and route the audio to a different computer (or standalone push).
And have the other computer have a low buffer size and use it only for vocals and guitar input.

Best regards,
Mangrove

Spikes and pops are a product of your CPU power, interface, and the plugins you're using. Certain instruments or devices introduce more pops. Even on my M2 Max, Superior Drummer loves to pop when I change the volume at a buffer below 512 as an example. I don't really have an issue with enabling devices or changing outputs, so one of the first 2 things I mentioned (or both) is likely the culprit there. As far as live performing goes, you have some options.

I deal with this 2 ways: First, I'll use Gig Performer and zero latency plugins. Essentially acts like UAD console where you can track with very minimal latency regardless of your session size/buffer. Setting it up to do "zero latency" recording is a bit dependent on your interface and its ability to "loopback" audio, but it was a great investment for me. I still use my Apogee Element for this since it has low latency, so my only real bottleneck is buffer size, which GigPerformer allows to go down to 16 samples and still works with this interface (it may not on others). This would be a much more efficient method than routing audio to a completely different machine. You'll just have to watch total CPU usage for all processes on your computer.

Second way is to just print everything in the track, save as a new session with only audio, be it 2-track or stems. That way you're only dealing with the latency of your interface driver since there's no plugins present in the session that can introduce more latency or eat up CPU. Turn off warping on your stems if you really want to eliminate your CPU usage, and track through plugins with little or no latency.

Your third option would be direct monitoring from your interface, recording the dry signal, and then processing after, but it sounds like you're trying to track through fx for live performance, so this is more of a "studio" solution.

Hope this helps!