ACE Studio voice synthesis

[jonljacobi]

Not sure this is the correct forum as Ace Studio is marginally a DAW and all about vocal synthesis (actually sample manipulation to mimic voices), but I have been messing around with it for a bit.

It's startling how effective it can be, but it has one major flaw that keeps me from using it. Actually, I don't love any of the voices i've tried, but...

The major flaw seems to be handling compound consonants at the end of a English word such as the "ltz" in "waltz". It allows you to enter the phonemes "l", "t", and "z", but it generally only renders the first. Or two if you're lucky as in "ts", but that's not a sure thing either.

I've tried about half a dozen voices and a couple are better than the others with this. But none are satisfactory in this regard.

The program is Asian in origin, and that's likely the issue when it comes to English. Most of the singers are from that part of the world. I don't know enough about Asian languages (a smattering of Mandarin) to say why, but until it can handle compound consonants, it's only marginally useful for English lyrics.

Let me know if I'm missing a trick, and what that might be. It's fascinating that the process is as good as it is regardless. I might try the custom voice but that requires payment. If they get back to me about my report, I might try that.

[jonljacobi]

Conversing with myself, I've managed to produce audible compound consonants by splitting them into separate notes that are very close in time. A pain, but it works "okay". I can now at least hear "ds", "tz", etc.

If your singing sucks, ACE Studio is worth a try. I'm going to pony up to try the custom voice (me) and I'll get back to you.

[jonljacobi]

As apparently, I'm the only one interested enough to play with it... The audio to MIDI, while not perfect, is light years better than what I get within Ableton Live. Live of course has no audio to lyrics at all and ACE Studio also translates those.

I can actually sing into it (pathetically I might add) and have a reasonable facsimile of what I intended (though much improved) be produced. It's quite remarkable.

[VitaminD]

I've seen at least one other thread on this here, if you're interested.

I'm not using it though because its a subscription only. I was initially really impressed with the quality. However, listening more, I think it needs another generation or two before it gets rid of the synthy-ness of the vocals, at least judging from the youtubers using the tool. Some can get close if they really work the controls, but even then there is still a synthetic or flatness to the voice models that 'rings' throughout the performances that I don't want as lead.

[Krakatau]

looks promising...

[Touch The Universe]

I'd like to test it but my bank declined the payment attempt.

Can I send somone with access to this program some lryics and a midi file to test this thing to see if its worth the trouble?

[jonljacobi]

It's a mixed bag. You can get what I consider really good "scratch" vocals. I found the original set of vocalists very generic and weak. Also, and I don't mean this as slander, as some will like it, but very pop Asian in nature. However, they just recently added some that I think have a lot more character, and there is always the custom voice which I've used to capture my own.

Note that on the custom voices, it helps to have samples covering several octaves. The character can change drastically when you switch using ACE if you don't.

It can take a lot of tweaking of the length of the consonants and vowels, as well as tweaking the energy, tension, air and other things to get pronunciation correct. I get the best results from creating the MIDI elsewhere and importing it. Then adding the lyrics. You can do this within the program but it tends to over-complicate things.

Put another way, it's extremely difficult to get the "soul" or energy that a good singer can bring. Maybe impossible for the rawer genres. I've yet to do it and I've gotten pretty facile with the editing. But it's really amazing how close you can get. For harmonies it sounds excellent.

[alienmachine]

I briefly tried it. It crashed so many times that I gave up. Definitely not ready for prime time.

[XComposer]

I'd be interested if only they could also sing in Italian and in Latin, beside English (working from scratch with phonemes only to imitate these other languages would not be practical.)

[jonljacobi]

Crashed on me once, but never since. But I'm on the Mac. I've never tried typing in Italian, but it does translate your lyrics into phonemes for you. It might do better than you'd think, but the language is not officially supported. Maybe you can work a deal with them.

[XComposer]

Crashed on me once, but never since. But I'm on the Mac. I've never tried typing in Italian, but it does translate your lyrics into phonemes for you. It might do better than you'd think, but the language is not officially supported. Maybe you can work a deal with them.

Interesting. It must be established whether it translates my typing into phonems according to the English pronunciation or according to a correct Italian pronunciation (the former would be useless). They are very different.
Anyway, it's not a priority for me. It would just be a help to simulate the result of a vocal composition before it really gets performed (and sung). Maybe I'll check it sometime in the future.

[dune_rave]

You can' t demo it "freely".
Wait, I will explain.
You have to pay for subscription (or buy the full "LIFETIME" version) to use this software. The subscription is billed yearly, and it's 200 USD if you subscribe. There's a money back guarantee, so you can get your money back if you are not satisfied.
So pay 200 USD to demo it!
I'm the only one, who think this is not fair?
For demoing, I would gladly pay 1 month price, that's around 17 BUCKS!
But a Whole year?
I don't want to "borrow" 200 bucks!