Cool new "spectral tools" in Bitwig

It’d be the fft size + the hop no? Either way, I’ve found for things like this (like the sweet sounding FrameLib spectral freeze) that at lower fft sizes the sound quality drops really quickly, so it becomes hard to find a usable spot where it’s fast enough to use, and good enough sounding to be worth using.

The latest version sounds real nice.

Something like that yes.

You mustn’t use any spectral processing in realtime then :laughing: You were born in the wrong era!

Hehe, if it’s >512, it’s too slow for me!

Not that it would make a massive difference, but I do look forward to an M2+ world where everyone runs at 192k, so I can shrink my I/O vector down and possible be able to do realtime stuff with (the equivalent of) 512 or, god forbid, 1024!

That being said, for some stuff it isn’t as critical. It’s just when there’s audible transients, it becomes super apparent, and the loudness splitting is definitely one of those cases.

(I also have a great sounding “faux spectral freeze” based on one of @pete’s reverb that has no latency, and is acting as my Framelib freeze replacement for the time being)

you won’t get more precision than 48k 128 samples… except in bat range :slight_smile:

How does the freeze work with no latency? Surely there must be some model of time built into it for projecting forward some frozen audio samples?

It’s more the roundtrip time I’ll save by having an I/O vector size or 16 @ 192k, so I’ll have a few ms to “spare” to try out chunkier fft sizes.

It’s not actually a spectral/frequency-based freeze, but a reverb-based process that you can freeze the state of the buffers (or however it works). So the moment you send some energy into the system, it’s already sounding. I actually did a bit of testing to try a hybrid approach ala multiconvolve where the first initial bit is the time-based “reverb” freeze and then that crossfades over to the Framelib one, but in the end I think the time-based one sounds good and usable on its own. It’s a bit since I did the testing, but I remember actually preferring the sound of the freezeverb approach on percussive stuff anyways.

It seems IRCAM decided to counter-attack:

Some new videos with more info/detail:

I love this guy’s tone and cadence

wooooow this is full-asmr-trigger-space!