Is it time for SUPER plugin?

[DJErmac]

I’m quite sure this was this one :
https://github.com/CorentinJ/Real-Time-Voice-Cloning
It’s the most advanced I’ve heard so far.

I still see no use for it excepted good jokes.

Nice! Thank you.

As Polki suggested, feeding it a voice and an instrument or an instrument and a texture might be a viable and welcome alternative to [regular] vocoding/noise-shaping. - I see it is based on vocoding.

I’m 99,9% convinced this example is indeed based on vocoding.
I read the whole thing at the time... I wanted to make jokes...

making jokes is serious business!

Finally someone who understands the real demand...

i guess we can come back in march and see if it happened because demand!

Hey, looks like this is the final conclusion of this topic actually, now that it’s officially planned for march 2023. We can all go back to our normal activities.

[whyterabbyt]

I’m quite sure this was this one :
https://github.com/CorentinJ/Real-Time-Voice-Cloning
It’s the most advanced I’ve heard so far.

Hmmm, an older one. No obvious samples, but the page there refers to their own closed-source product (resemble.ai) as better.

altered.ai is the smoothest-sounding one Ive seen used in realtime, and the voice copying is good, but not 100% identical.

[vurt]

I’m quite sure this was this one :
https://github.com/CorentinJ/Real-Time-Voice-Cloning
It’s the most advanced I’ve heard so far.

I still see no use for it excepted good jokes.

Nice! Thank you.

As Polki suggested, feeding it a voice and an instrument or an instrument and a texture might be a viable and welcome alternative to [regular] vocoding/noise-shaping. - I see it is based on vocoding.

I’m 99,9% convinced this example is indeed based on vocoding.
I read the whole thing at the time... I wanted to make jokes...

making jokes is serious business!

Finally someone who understands the real demand...

i guess we can come back in march and see if it happened because demand!

Hey, looks like this is the final conclusion of this topic actually, now that it’s officially planned for march 2023. We can all go back to our normal activities.

well that's me off to not do something im supposed to do.
or do something im not supposed to do.
depends on how i feel when i don't go anywhere to do anything.

[Leo1999]

Wouldn't be surprised to see the "sing like x" thing done fairly well in the next few years. Spoken word audio deepfakes are getting pretty good and it's a natural extension of that. I'm not convinced the rest of it makes any sense at all. If you want to play guitar like Jimi Hendrix with minimal playing capabilities, the thing needs to totally transform what you're doing. Your timing, articulation, and note choices are going to need to be changed completely. Are you just supposed to strum chords and the AI does the rest? At that point why not cut out the middleman or -woman and just enter chords and other instructions in a UI? Not unlike the prompts from the current generation of AI image generators, actually. Transforming drum playing makes even less sense because there's not even chords involved.

Also as an aside I'd much rather see AI mixing and mastering plugins that actually work well. Let me keep doing the fun stuff and let the computer handle all the boring technical stuff.

iYou don't have to be a plugin dev to see that it's simple and there could be several options to do it. Rhythm guitar, how do you want it to be played?
You feed AI with all songs of Hendrix etc and the plugin knows how Jimi etc plays rhythm guitar.
1st option
You tell the plugin the chords, A, D and E major and if you know some guitarists you already hear what it sounds like when Hendrix, Page, Angus Young, Clapton would play it, maybe you could also select something like Blues, Hard Rock, Ballad etc. Simple!
The plugin knows all Hendrix songs and would take the average Hendrix rhythm guitar, if you want Hendrix and maybe also the average Hendrix sound

2nd option
You could refer to a song, play rhythm like spanish castle magic, or smoke on the water or under the bridge etc. simple! It would always refer to your performance. AI would then refer to the Hendrix performance that is the closest to yours and apply it.

3rd option there are certain rhythm techniques or styles I do not know the correct terms
Under the bridge / intro Little wing is a special style
Wild thing is another one
There is a house in New Orleans is a completely different one,
just select the rhythm style you want

I watched a drummer on yt, he talked about drummers from 1950s to
1990ies and showed their styles.
If a real drummer can do it, a plugin could do it, too.
Let's take Hendrix, his drummer Mitch Mitchell played like no
other drummer, you want your drums to sound like Mitch,
Super plugin should be able to do it

[DJErmac]

I’m quite sure this was this one :
https://github.com/CorentinJ/Real-Time-Voice-Cloning
It’s the most advanced I’ve heard so far.

I still see no use for it excepted good jokes.

Nice! Thank you.

As Polki suggested, feeding it a voice and an instrument or an instrument and a texture might be a viable and welcome alternative to [regular] vocoding/noise-shaping. - I see it is based on vocoding.

I’m 99,9% convinced this example is indeed based on vocoding.
I read the whole thing at the time... I wanted to make jokes...

making jokes is serious business!

Finally someone who understands the real demand...

i guess we can come back in march and see if it happened because demand!

Hey, looks like this is the final conclusion of this topic actually, now that it’s officially planned for march 2023. We can all go back to our normal activities.

well that's me off to not do something im supposed to do.
or do something im not supposed to do.
depends on how i feel when i don't go anywhere to do anything.

I understood absolutely nothing but I enjoyed it a lot.

[Leo1999]

Wouldn't be surprised to see the "sing like x" thing done fairly well in the next few years. Spoken word audio deepfakes are getting pretty good and it's a natural extension of that. I'm not convinced the rest of it makes any sense at all. If you want to play guitar like Jimi Hendrix with minimal playing capabilities, the thing needs to totally transform what you're doing. Your timing, articulation, and note choices are going to need to be changed completely. Are you just supposed to strum chords and the AI does the rest? At that point why not cut out the middleman or -woman and just enter chords and other instructions in a UI? Not unlike the prompts from the current generation of AI image generators, actually. Transforming drum playing makes even less sense because there's not even chords involved.

Also as an aside I'd much rather see AI mixing and mastering plugins that actually work well. Let me keep doing the fun stuff and let the computer handle all the boring technical stuff.

You don't have to be a plugin dev to see that it's simple and there could be several options to do it. Rhythm guitar, how do you want it to be played?
You feed AI with all songs of Hendrix etc and the plugin knows how Jimi etc plays rhythm guitar.
1st option
You tell the plugin the chords, A, D and E major and if you know some guitarists you already hear what it sounds like when Hendrix, Page, Angus Young, Clapton would play it, maybe you could also select something like Blues, Hard Rock, Ballad etc. Simple!
The plugin knows all Hendrix songs and would take the average Hendrix rhythm guitar, if you want Hendrix and maybe also the average Hendrix sound.
AI would then refer to the Hendrix performance that is the closest to yours and apply it.
It always refers to you performance

2nd option
You could refer to a song, play rhythm like spanish castle magic, or smoke on the water or under the bridge etc. simple! It would always refer to your performance.

3rd option there are certain rhythm techniques or styles I do not know the correct terms
Under the bridge / intro Little wing is a special style
Wild thing is another one
There is a house in New Orleans is a completely different one,
just select the rhythm style you want

I watched a drummer on yt, he talked about drummers from 1950s to
1990ies and showed their styles.
If a real drummer can do it, a plugin could do it, too.
Let's take Hendrix, his drummer Mitch Mitchell played like no
other drummer, you want your drums to sound like Mitch,
Super plugin should be able to do it

[_leras]

nobody enjoys microwave meals, we eat because we must at a given time or we die. therefore it's essential to fit a meal in. even if it means eating a tasteless pile of shite.

music, is not so essential. we can choose not to listen when we don't have the time.

why would you choose to listen to shite?

F##k. Make sure you don't get your afternoon tea ten minutes late. Dead huh?

I choose to listen to loads of different types of music. Some of it is a bit shite, but still a bit enjoyable.

[Schnickschnack]

ugh

[JockMcRonan]

outta here

[thecontrolcentre]

Yup

[Leo1999]

...just a few minutes ago
I could read the post of a professional mixing and mastering engineer
and he writes artists will be replaced by AI.

"Who should be scared of AI are the artists,
AI takes over, there is no more artistic careers
for artists, zero, none"
Times are changing!
But you could also see positive things,
AI is not bad only!
I read so many times that the result is important,
no one cares whether you used this or that,
who cares if a guitar part was enhanced or adapted
to the style of a famous guitarist by a plugin.
You don't play guitar very well?
No one cares if the result sounds good!

[jamcat]

I see, then you might know the answer

what, this lot?

i think he's second from the left.

This song gave me ADHD.

[briefcasemanx]

I think it's time for super plugin

[DJErmac]

I think it’s time for all musicians in the world to stop working and learn origami instead to pass time, as we’re all doomed by super plugin to be totally useless. This is destiny. The prophecy has risen from Hell. Super plugin killed us all today and we don’t even realize it yet, as it will be released in march.

[Etienne1973]

One day AI might replicate the universe.