KVR Audio

rodyy · Post by **rodyy** » Mon Nov 13, 2017 5:24 am

I’ve got a pretty bad singing voice, so 2 years ago I decided to build a plugin that can sing.
Here are a few early examples.
https://soundcloud.com/emvoice

I’d love to get some feedback, see what features you guys would like us to build. Btw we’ll release a beta to about 50 testers in a couple of months, PM me if interested! (Mac and good production skills required)

V0RT3X · Post by **V0RT3X** » Mon Nov 13, 2017 5:31 am

Neat sounding plugin.

If it's priced reasonably I might check it out as a sound design tool. I'd never use it to replace a real singer though..

rodyy · Post by **rodyy** » Mon Nov 13, 2017 9:06 am

Thanks! Yes, the point is not necessarily to replace a real singer. With the next update of the engine we will get a bit closer to a human voice though

kelvyn · Post by **kelvyn** » Mon Nov 13, 2017 10:01 am

Sounds pretty good already. Pretty interested what one can do with it out of the normal pop context. I would be very interested in beta testing when you are at that stage.

Astralv · Post by **Astralv** » Tue Nov 14, 2017 3:00 am

Hey there

Just lost the reply I was typing, have to retype... Congratulation in taking huge step in creating artificial vocalist. I can envision our future where we would not have to worrying about hiring a vocalist and will be able to just buy software with the voice we like. Great job!

I dont know how this plugin working but wondering why it has certain personality? Disclaimer: I am a woman and saying it from woman's perspective. Your female model talks excessively sexual. Women don't normally talk like this- this is stereotypically overemphasized. At the same time- your male model sounds like a nerdy boy next door. I cant think of a style of music that would benefit from vocal like this. May be documentary about troubled teenagers struggling with acne... Your male model is absolutely not sexual. It maybe cute but definitely not attractive. I noted in first demo you going for Depeche Mode. Great idea. The difference is that your model sounds nasal and short phrases and words make it sound immature. It needs to be less "Spider-Man" and "Harry Potter" sounding, less Christian singer songwriter, more underground, sensual and erotic. How do you do it? I dont know. But if you could get rid of nasal and fix short words and make voice to actually hold the notes and make it few years older (Depeche Mode David Gahan was born in 1962), it may sound better. Otherwise- very inspiring. Thanks.

rodyy · Post by **rodyy** » Tue Nov 14, 2017 4:34 am

The plugin is based on actual recordings of a singer. So the voices are a reflection of the original performance we capture, not the result of our own choices!
We're looking for vocal talent (especially female singers) to create new voices btw.

rodyy · Post by **rodyy** » Tue Nov 14, 2017 6:57 am

kelvyn wrote:Sounds pretty good already. Pretty interested what one can do with it out of the normal pop context. I would be very interested in beta testing when you are at that stage.

We will release a private beta in about 2 months

HiEnergy · Post by **HiEnergy** » Tue Nov 14, 2017 9:44 am

What a pity I have no Mac but "merely" a Windows PC... otherwise I'd apply for beta-testing this in a heartbeat.

rodyy · Post by **rodyy** » Tue Nov 14, 2017 1:42 pm

Yes, we're Mac users so we're starting to test on Mac. But the PC version should follow shortly. Thanks!

Astralv · Post by **Astralv** » Wed Nov 15, 2017 4:21 am

I am listening Ethera while I am typing this- wish you had those low female voices. http://www.timespace.com/product/ETHBUN ... ad%29.html

How would you record a talent- you need someone locally, right?

PC version, please! Dont know what you still doing on Macs- this is so yesterday! Time to build your own computers!

Astralv · Post by **Astralv** » Wed Nov 15, 2017 4:43 am

By the way- I suspected- it was your vocal in demos- is it? It really not that bad, just need to practive with other vocalists and work on holding the notes.

melomood · Post by **melomood** » Wed Nov 15, 2017 6:34 am

Astralv wrote:Hey there

Just lost the reply I was typing, have to retype... Congratulation in taking huge step in creating artificial vocalist. I can envision our future where we would not have to worrying about hiring a vocalist and will be able to just buy software with the voice we like. Great job!

I dont know how this plugin working but wondering why it has certain personality? Disclaimer: I am a woman and saying it from woman's perspective. Your female model talks excessively sexual. Women don't normally talk like this- this is stereotypically overemphasized. At the same time- your male model sounds like a nerdy boy next door. I cant think of a style of music that would benefit from vocal like this. May be documentary about troubled teenagers struggling with acne... Your male model is absolutely not sexual. It maybe cute but definitely not attractive. I noted in first demo you going for Depeche Mode. Great idea. The difference is that your model sounds nasal and short phrases and words make it sound immature. It needs to be less "Spider-Man" and "Harry Potter" sounding, less Christian singer songwriter, more underground, sensual and erotic. How do you do it? I dont know. But if you could get rid of nasal and fix short words and make voice to actually hold the notes and make it few years older (Depeche Mode David Gahan was born in 1962), it may sound better. Otherwise- very inspiring. Thanks.

So basically you think the male singer sounds like The Postal Service

rodyy · Post by **rodyy** » Wed Nov 15, 2017 7:46 am

Astralv wrote:By the way- I suspected- it was your vocal in demos- is it? It really not that bad, just need to practive with other vocalists and work on holding the notes.

Not my vocals

We can work with vocalists anywhere, as long as they can record themselves.

FLWrd · Post by **FLWrd** » Wed Nov 15, 2017 7:41 pm

It's not bad, but there are some giveaways: Julian's vowels sound like single-cycle loops. A bit of vibrato or morphing between two versions of the same vowel might help.

"Take me away" sounds quite impressive, though. The voice sounds processed (telephone-style EQ, heavy pitch correction going on vocoding), but acceptable, and the delivery is believable in the psychedelic context. As was remarked above, it's not a full-bodied voice.

I am interested in a beta test, so if you need testers (still on OSX 10.12, Logic 10.3).

rodyy · Post by **rodyy** » Thu Nov 16, 2017 1:32 am

Thank you for the feedback! "Take Me Away" is our original proof of concept, I created the vocals with a standard text-to-speech engine and pitch-shifted the output to make it sing. In the context of this song, and with all the added production, this voice works indeed quite well.
The other demos (Julian) are the actual output of our plugin. While I tried to make them sing "nicely", the songs I chose may not be the perfect fit for this voice - and they're a cappella, which is pretty unforgiving even for a real singer. PM'd you

Feedback on a singing voice generator (plugin)