State of the art AI audio stem splitting and processing for artists/producers, forensics, karaoke. Join us https://nitter.cf/t.co/UzVUSQjt0j
Sweden
Joined November 2019
- Tweets3.4K
- Following249
- Followers2.2K
- Likes1.8K
We’ve always been local-first.
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
When we cook, we cook slow, but also
when we cook, we cook good!
What's a 5 stem, 10 stem or 16 stem model to a 56 stem model? 😎
Coming soon!
If you you work with audio, you don't want to miss this!
What's coming up in the next major update.
Follow. Like. Share.
Stay tuned!
One last thing for today.
Showcasing ASTRA+ with Focus & Details.
Comparing STFT with ASTRA+ with F&D in the video.
Insane!
iZotope RX12, Steinberg Spectralay, Acon Digital, and others = 💀
All the DSP algorithms aside, our ASTRA+ renderer is superior.
Audio editing will not be the same after
we drop SXA (Splitter Studio X Audio).
This is the first time I share this to the public.
You'll find all the usual processing modules that you'll find in other audio editing apps, but BETTER!
Why better? Well, because most of our modules are REALTIME!!!
No need to process and wait to hear the results. Just open the module and get to listening.
And...
The name ASTRA came about as an easy-to-say acronym of "Advanced Spectral Transformed Resolution Algorithm".
ASTRA gives a SIGNIFICANTLY better spectrogram than traditional methods, and ASTRA+ pushes it just a tiny bit more.
Why is this needed?
When we first started ramping up our AI efforts back in 2021, we needed a better way of letting the AI know what it was looking at, and a sharper spectrogram would yield better output.
We researched this topic and came up with ASTRA.
Idk why I'm being all secretive. Here is one of the projects we've been working on.
We call it ASTRA and its successor ASTRA+, and no, we did not copy OpenAI.
So what is it?
