A stereo image production trick
Вставка
- Опубліковано 29 лют 2024
- A little trick for widening a track using Combobulator by DataMind Audio.
Watch the full tutorial in the HCA Feed (if you aren't a Hardcore Abletoneer yet, subscribe today - mrbillstunes.com/landing/hard....
Mr bill never fails the fill my space with information ❤
Pretty sick! I know it's brand new, but I'd like to see more videos about this soon because the last thing I need is to buy more plugins !LOL
I have a few different techniques, but it depends on the impact you're trying to have
I usually do a few layers of sides and one or two middles.
Sometimes i even slightly detune my sides from my middle :)
I should send you some of the music i do. You'd probably like it! Been on a tipper trip lately.
Sounds like a side chain pumping effect is being created?
i didnt get it, but i will understand this
i used to have respect for you
When you solo'd just the side information, it's doing some gnarly out of phase thing. It's not as noticeable when you bring the middle channel back in though. I think combobulator's side image thing may be being generated similar to how Wider works with each side being comb filtered maybe? I remember a while back you mentioning in a podcast you wanted to take a mono signal and have AI generate truly new side-image information... but is this really doing that?
The side information is literally the difference between the left and right channel, it's always going to be out of phase when it's solo'd
@@LarsWilms If it was truly different audio info it shouldn’t be. Also there’s no L/R, it’s just a mono signal.
they certainly seem to claim it is, but their website is sparse on details and it's not clear to me if the width is an actual parameter of the model or is being done via more traditional DSP methods on the model's output signal like you're suggesting.
Edit: I found the user manual and this is what it says
"Width is just like Offset, except it applies a gentle offset to the latents used to synthesize the left vs right channel. More difference in the latents between the left and right channel means a more variance between these channels, resulting in a wider stereo image."
So seems like it's not a DSP trick and it is indeed generating the left and right channels from different places within the model's latent space giving you a slightly different timbre for each channel.
@@iswm What accounts for the out of phase sound when the sides are solo’d in the video? Is the plugin generating sounds deterministically and offsetting them sample-wise for the stereo image?
@@PaulEubanks I couldn't tell you. My best understanding is that it's running 2 channels through the model to resynthesize outputs, whether or not random seeds are involved I have no clue, but when increasing the width it slightly shifts the channels apart in latent space causing it to resynthesize each channel slightly differently, and that slight variance in the resulting timbre of each channel causes a natural stereo width effect that becomes more pronounced as the difference between the channels increases. I'm assuming the difference in timbre between the channels is quite subtle and the two channels are still similar enough that it can still produce phasing issues, but I'm completely speculating here.