Intel AI effects for Audacity

[agp]

Introducing OpenVINO AI effects for Audacity
www.audacityteam.org/blog/openvino-ai-effects/?ref

The effects are:
Noise Suppression tool.
Transcription tool powered by OpenAI's Whisper.cpp.
Music Generation, and Music Style Remix based on Stable Diffusion to generate music based on prompts.
Music Separation to split a song into its vocal and instrumental parts.

I can finally ditch Moses on android to separate the instrument tracks!

[Rodney]

Damn, that's great! AI is really a useful Tool. Thanks for sharing!

[oddy.o.lynx]

I'v built this successfully. One thing though, when installed in

Code: Select all

/usr/local

it had an ioctl error and would not load openvino. Putting it under

Code: Select all

~/opt/audacity

the plugin loads.

Haven't tested it yet but will soon.

[Largos]

What is this all useful for?

Noise Suppression: There is already lots of stuff for that available.
Transcription Tool: I guess this is for video subtitles
Music Generation tool: ??
Stem Separation: People who do sample based music seem excited by this tech but there are always nasty artifacts all over the separated parts.

[d.healey]

For subtitle generation (for video) I use the OpenAI tool within Kdenlive, which also has translation support.

[funkmuscle]

What is this all useful for?

Noise Suppression: There is already lots of stuff for that available.
Transcription Tool: I guess this is for video subtitles
Music Generation tool: ??
Stem Separation: People who do sample based music seem excited by this tech but there are always nasty artifacts all over the separated parts.

Who knows but noise suppression, I think that's a bonus even though we have a lot because if this one could actually pull the annoying frequencies out without changing the tonality of the instrument or vocal then that's huge.

Try removing the bleed from say the snare drum of a live recording. No matter what tricks you use you still will hear a little bit of that high hat if it was hit at the same time as the snare drum and sometimes you want to boost that frequency that gives that extra crack but at the same time you're boosting that annoying hi-hat frequency so if an intelligent noise suppression plug-in can actually find that frequency only and subtract it I think that a godsend.

All the other stuff though man I'm with you on that because I'd like somebody to explain what is it good for outside of speech because I think that's how they are promoting it.

My only care was/is is a very intelligent or AI noise suppression plugin because the two big platforms have tons.

Tired of going online to look for ways to get rid of that excess bleed to only have some guy on YouTube who's being sponsored by some big label or big company sorry saying he's using the latest this and that and listen to the perfection and I just have to keep reminding myself that Linux at its current stage for pro audio is like what the guys in the '70s and '80s had to put up with so in the long run, it just makes us better at what we're doing because we have to use old school techniques but still it would be so nice when you're in a rush to get something done to be able to just have some kind of noise suppression plugin to just clean up all of the nasties.