Voices for cloud-based singing synthesizer Emvoice One are half-off.

[Fleer]

There’s a hole, dear Liza

[sarine]

21 hours ago, Bapu said:

ircam trax v3 has the ability to take a male voice and transform into female (or vice versa) based on source and result type.

Tis expensive ($399). I've used it with moderate success. Of course you can't expect lead vocal quality but 'sit the mix' bg vocals are quite acceptable. 

You can try it for free.

Flux IRCAM Trax is very good. It still needs a lot of attending, but that's just how it is with these pitch/formant transformers without AI smeared all over them yet (I'm not complaining).

If this computationally complex step (that is difficult to tune in) is part of your routine, you need to flip your perspective so that instead of doing your thing and expecting technology to fix it, you get familiar with the technology so that you can do your thing in a way that respects its limitations while making it easier for the algorithms to do their thing. (As a sidenote; It's similar to how if you're designing a 3d model to be printed using a FFF/FDM process, your whole CAD process is subordinate to the mechanics of the printing process.)

Even if you can't sing the correct pitches, you want to have as clean a source recording as possible (staying on scale has very little to do with that), so pay attention to all the extra noises you're making and get rid of everything you can before it even hits the microphone, and clean the rest before doing more involved pitch/formant processing.

It's possible that the voice transformer just doesn't like your natural voice, but may be pleased with you singing in a voice that feels slightly unnatural, producing more satisfactory results after all processing is done, even if the source sounds more awful to you. If you want to change the gender or age of your voice, you have to take it into account at the recording step, and you need to play with your voice and the processing software in parallel to find a co-operative zone where you can meet in the middle. When you find that zone, hone that voice and save those settings, and don't expect it to magically work if your target age/gender or other impressions of the voice change - you're better off doing all the work for any specific voice separately (if you want good results).

Also, don't obsess over perfecting anything, usually it's a waste of time and even on "succeeding" it'll suck life & soul out of the voice - unless that's what you aim for, which is a totally appropriate target considering most pop music doesn't seem to appreciate the human element in sound anymore. 

[Reid Rosefelt]

I sang with all the bands I was in as a lead guitarist during my teens, and for a year or two I even was the lead singer in a band and didn't have to play.  But now... not so good. 

A few years ago I had a friend sing one of my songs, as he has a much better voice.  It took forever.  He didn't sing it the way I heard it in my head so I made him do take after take and then I did a lot of cutting and pasting.   The lesson I took from that is that I have to do the vocals myself and fix them as best I can. 

But I do stare at Emvoice a lot.  I'm really curious about it.  I have too many toys to play with, but it looks like it would be fun to have it sing silly things.  A hundred bucks isn't bad for a girl singer, even one that sounds a bit like a machine. 

[TheMaartian]

On 1/8/2022 at 9:10 PM, bitflipper said:

I once needed an angelic chorus in a song, so I recruited two of my teenage granddaughters. They were rather reluctant, though, as neither of them considered themselves singers.

However, once I showed them what Melodyne could do, plus the magic of double-tracking, EQ and reverb, their confidence was greatly boosted and in the end we got some really nice results.

As with Gary Numan and Persia, one of his three daughters.