OMG: Synth V choir

[ajay_m]

I would think every single vendor of choir libraries will be having some bargain sales, because this new Synth V offering seems to be an amazing breakthrough - a choir (three in fact) that can actually sing anything you want them to. And listening to the demos, the results are quite remarkable. I already have Synth V and while you do see a little of the blurring of lead consonants on some of the examples, with a bit of careful programming it's usually possible in my experience to improve that with some vocal editing.

It appears with VAT that the bundle price of USD299 for all three corresponds to £277. (and you will need Synth V at I think around £80 if you don't already have it). Or USD149 each.

Each of the three choirs have their charms, it's very hard to pick just one. I must admit this is tempting. It's kind of the last frontier really, being able to do this, would have seemed like science fiction a few years ago. This must be the best thing since sliced bread for choirmasters, since you can create a full demo of a piece then easily play the individual SATB parts, or give them to the choristers to practice.

Anyway, curious as to what folks think, listening to the demos.

[BJG145]

I remember trying this with Symphonic Choirs 20 years ago. It was pretty hard work. Fun though.

The section Multis really come into play with the third element of the overall package, Word Builder. This sits between the normal MIDI data input (the melody of the line to be sung) and the Kompakt Multi. Each of the special Multis is arranged across five MIDI channels and contains a combination of the vowel and consonant samples. Given the user-specified phrase, Word Builder then splits the MIDI input across the five channels of the Kompakt Multi to sequence the various vowel and consonant sounds together to create the required phrase.

https://www.soundonsound.com/reviews/ea ... nic-choirs

I expect things have come on a bit since then...will check it out. (Is this AI?)

[RichardT]

I'm hopeful for the future rather than impressed with what they currently do.

I wasn't won over by the classical choir demo - and I've listened to another demo on the Dorico forums with the same result, though it is better, I think.

https://forums.steinberg.net/t/synthv-c ... nt/1021017

At the moment, the voices don't flow properly from note to note IMO and they are in the uncanny valley some of the time.

For testing whether the music has emotional impact, I still prefer Noteperformer 4.5 with Hollywood Choirs in Dorico, even though they just sing vowels.

But synth V choirs are already great for seeing how the words work across a number of choral parts.

[OneWorld]

I would think every single vendor of choir libraries will be having some bargain sales, because this new Synth V offering seems to be an amazing breakthrough - a choir (three in fact) that can actually sing anything you want them to. And listening to the demos, the results are quite remarkable. I already have Synth V and while you do see a little of the blurring of lead consonants on some of the examples, with a bit of careful programming it's usually possible in my experience to improve that with some vocal editing.

It appears with VAT that the bundle price of USD299 for all three corresponds to £277. (and you will need Synth V at I think around £80 if you don't already have it). Or USD149 each.

Each of the three choirs have their charms, it's very hard to pick just one. I must admit this is tempting. It's kind of the last frontier really, being able to do this, would have seemed like science fiction a few years ago. This must be the best thing since sliced bread for choirmasters, since you can create a full demo of a piece then easily play the individual SATB parts, or give them to the choristers to practice.

Anyway, curious as to what folks think, listening to the demos.

Going off at a slight tangent. How you getting along with Synth V? I am looking around for a virtual vocalist to buy. I upgraded to Cubase 15 because it includes OmniVocal, although it is in beta form. It is OK, but either of the two singers only has, realistically, a range of about one octave, and that centres around C2 to C3, more of less.

OK I cannot realistically expect the voice of Pavarotti, and I am expecting a judicious listener to say "That's AI" yep I can live with that, as long as the results are acceptable, rather than distracting.

I have also been looking at IK Multi-Media's Re-Sing, however many people on YouTube claim that IK's Media Manager fills your hard drive with clutter and is too intrusive. I once used IK products and took them off for the same reasons.

I keep being pointed towards Suno, but that again comes with its irritations. I think what purveyors of such products should give a potential customer a free to use, unrestricted period of time, say 7 days to fully evaluate the product, after all one might be considering spending several £100's and that 'free to use for 7 days' offering should not sprinkle all sorts of stuff over the PC with marketing claptrap, once the evaluation period is over, your pc should be purge of all that is connected with what the product is, and an option to be removed from the seller's database so one doesn't get festooned with spam.

SOS Editors - considering there is an ever growing market for such products, is there an opportunity for a group test article in SOS?

As a slight aside, it was reported the other day, more and more, people are finding that products such as ChatGPT are far from the ideal and some 30% of users say they have experienced use where the output is a load of tosh - no surprises there then, those of use that aren't so easily impressed didn't expect anything else.

That's not to say it's all a load of rubbish, but like any technology we take it for what it is, and aid, not a substitute. In an ideal world I'd like a competent singer at my disposal, yes and the people in hell would like ice with their gin and tonic, and they have more chance of their being met than me singing in tune

[OneWorld]

I would think every single vendor of choir libraries will be having some bargain sales, because this new Synth V offering seems to be an amazing breakthrough - a choir (three in fact) that can actually sing anything you want them to. And listening to the demos, the results are quite remarkable. I already have Synth V and while you do see a little of the blurring of lead consonants on some of the examples, with a bit of careful programming it's usually possible in my experience to improve that with some vocal editing.

It appears with VAT that the bundle price of USD299 for all three corresponds to £277. (and you will need Synth V at I think around £80 if you don't already have it). Or USD149 each.

Each of the three choirs have their charms, it's very hard to pick just one. I must admit this is tempting. It's kind of the last frontier really, being able to do this, would have seemed like science fiction a few years ago. This must be the best thing since sliced bread for choirmasters, since you can create a full demo of a piece then easily play the individual SATB parts, or give them to the choristers to practice.

Anyway, curious as to what folks think, listening to the demos.

Going off at a slight tangent. How you getting along with Synth V? I am looking around for a virtual vocalist to buy. I upgraded to Cubase 15 because it includes OmniVocal, although it is in beta form. It is OK, but either of the two singers only has, realistically, a range of about one octave, and that centres around C2 to C3, more of less.

OK I cannot realistically expect the voice of Pavarotti, and I am expecting a judicious listener to say "That's AI" yep I can live with that, as long as the results are acceptable, rather than distracting.

I have also been looking at IK Multi-Media's Re-Sing, however many people on YouTube claim that IK's Media Manager fills your hard drive with clutter and is too intrusive. I once used IK products and took them off for the same reasons.

I keep being pointed towards Suno, but that again comes with its irritations. I think what purveyors of such products should give a potential customer a free to use, unrestricted period of time, say 7 days to fully evaluate the product, after all one might be considering spending several £100's and that 'free to use for 7 days' offering should not sprinkle all sorts of stuff over the PC with marketing claptrap, once the evaluation period is over, your pc should be purge of all that is connected with what the product is, and an option to be removed from the seller's database so one doesn't get festooned with spam.

SOS Editors - considering there is an ever growing market for such products, is there an opportunity for a group test article in SOS?

As a slight aside, it was reported the other day, more and more, people are finding that products such as ChatGPT are far from the ideal and some 30% of users say they have experienced use where the output is a load of tosh - no surprises there then, those of use that aren't so easily impressed didn't expect anything else.

That's not to say it's all a load of rubbish, but like any technology we take it for what it is, and aid, not a substitute. In an ideal world I'd like a competent singer at my disposal, yes and the people in hell would like ice with their gin and tonic, and they have more chance of their being met than me singing in tune

As I write this I am listening to Taichovsky's Italian Concerto played by the Boston Symphony Orchestra on BBC Radio 3, by gum that chap knew how to know out a tune or tune or two.

[FrankF]

I would think every single vendor of choir libraries will be having some bargain sales, because this new Synth V offering seems to be an amazing breakthrough - a choir (three in fact) that can actually sing anything you want them to. And listening to the demos, the results are quite remarkable. I already have Synth V and while you do see a little of the blurring of lead consonants on some of the examples, with a bit of careful programming it's usually possible in my experience to improve that with some vocal editing.

It appears with VAT that the bundle price of USD299 for all three corresponds to £277. (and you will need Synth V at I think around £80 if you don't already have it). Or USD149 each.

Each of the three choirs have their charms, it's very hard to pick just one. I must admit this is tempting. It's kind of the last frontier really, being able to do this, would have seemed like science fiction a few years ago. This must be the best thing since sliced bread for choirmasters, since you can create a full demo of a piece then easily play the individual SATB parts, or give them to the choristers to practice.

Anyway, curious as to what folks think, listening to the demos.

Any links we can click?

[OneWorld]

There was an advert the other week which went straight to a demo, I have lost the link now but here's one on YouTube, jump straight to 20:20 as per usual with YouTube vids the presenter, after the obligatory "Wassup Guys Wassup" spends half the time blathering on.

https://www.youtube.com/watch?v=7rDqgGZOuEU

It is quite impressive and you can edit any of the 16 voices, very useful

[ajay_m]

Synth V itself is pretty good I think - it's reasonably priced and some of the vocalists (Solaria is generally I think the one to get initially) are very good.

You do need to put quite a bit of work into creating a completely realistic vocal line because of course real vocalists glide around in pitch, take breaths, do all kinds of things. Also to get initial consonants to sound sufficiently forceful sometimes takes a bit of effort. But the end result can be quite impressive e.g here Solaria is singing "All By Myself" https://www.youtube.com/watch?v=Xl2w01UGaMY

The only irritation for me is that I hate dark UI themes but you can't change to a light background with dark text. In fairness, the UI is scaleable (looking at you NI!) and quite legible, this is just my own personal peeve. A lot of people love dark themes, so, hey...

A streamdeck can be helpful as you can assign a whole bunch of editing options to buttons which makes the process of editing a vocal line a lot less tedious, but it is still a fair bit of work - however, if you don't have a vocalist, you can sure put together a damn good demo and get the rest of your mix to sit with the vocals and then look at putting live vocals in when the track is pretty much sorted. Also, it's a lot of fun having Solaria say terribly rude things about certain people in this world, right now, I have to say

[Slipkid]

I use synth v as an ara plug in. I ‘sing” the parts which gives me the phrasing that I want. My pitching isn’t terrible but I just hate the tone of my voice.
Once I get a take of my voice that is acceptable (perhaps with a little help from melodyn - which I then render) I duplicate that track and add synth v to one version and melodyne to the other version. The reason I do this is that if you add synth v in ara mode to your vocal track your original audio disappears and is replaced by the synth v engine.
Using the (can’t remember the exact terminology I’m away from home at moment) extract note info / use phonetic interpretation on synth v I create a vocal track . I have to say the tracking/interpretation can be anywhere between bloody impressive and bloody hilarious. However, tweaking it takes no time in most cases especially with the phoneme stretch control in the latest version.
Sometimes I find there are little issues with phrasing /pitching and that’s where having a melodyne copy comes in really handy .
I have several voices. My go to is now Liam as “he’ seems more English than American in pronunciation /accent than some of the others. However I like Havox and Nine-0/for heavier stuff and Kevin is useful. Saros I regret buying and Sheena although I got “her” free is of limited use. Solaria is excellent as is Natalie.
All in all it’s an impressive bit of kit I thorough recommend it.

John

[James Perrett]

But the end result can be quite impressive e.g here Solaria is singing "All By Myself" https://www.youtube.com/watch?v=Xl2w01UGaMY

That's not impressive at all. There are glaring issues with that vocal and it is very obviously AI.

I would urge anyone thinking about using these tools to remember that the voice is the most important element of a song and that people are very used to hearing the human voice. A robot voice might be fine for a rough demo but, if that's the current state of the art, then it isn't going to be good enough for a commercial recording - unless you are working in a genre where robotic sounding voices are accepted.

[RichardT]

But the end result can be quite impressive e.g here Solaria is singing "All By Myself" https://www.youtube.com/watch?v=Xl2w01UGaMY

That's not impressive at all. There are glaring issues with that vocal and it is very obviously AI.

I would urge anyone thinking about using these tools to remember that the voice is the most important element of a song and that people are very used to hearing the human voice. A robot voice might be fine for a rough demo but, if that's the current state of the art, then it isn't going to be good enough for a commercial recording - unless you are working in a genre where robotic sounding voices are accepted.

I feel the same way about AI voices. They are just not there yet - I don’t even use them in choir mock-ups yet, let alone music to release.

However, in a few years things may be different.

[Arpangel]

But the end result can be quite impressive e.g here Solaria is singing "All By Myself" https://www.youtube.com/watch?v=Xl2w01UGaMY

That's not impressive at all. There are glaring issues with that vocal and it is very obviously AI.

I would urge anyone thinking about using these tools to remember that the voice is the most important element of a song and that people are very used to hearing the human voice. A robot voice might be fine for a rough demo but, if that's the current state of the art, then it isn't going to be good enough for a commercial recording - unless you are working in a genre where robotic sounding voices are accepted.

I’ve heard Synth 5 and a friend did some vocal things with it, totally unimpressed, it was not good, I don’t know why people get so excited about it.

[ajay_m]

You may want to look back - and smile- as Herbie Hancock uses a vocoder so he can sing.
Technology has come a long way and, yes, it's not perfect. But this lets you write songs and lyrics and experiment with melody lines and if need be, you can do it all with a pair of headphones on and not disturb the neighbours.
And in any case a lot of modern auto tuned real vocals already sound unnatural.
So I think it's an interesting and useful product, as part of a creative toolset, for sure. And if I were to make a bet that someone will produce a hit track using this technology, I would not be surprised to see that happen within the next year or two.

[Hugh Robjohns]

I remember when if you wanted the sound of strings and couldn't afford the real thing, your choices were limited to a Solina (and its peers) or a Mellotron.

Things have come a very long way in the artificial strings department since then.

Choir V is by far the best and most versatile fake choir I have heard to date. If anyone knows anything better I'm all ears...

Sure, there are glaring errors and unrealistic elements, currently... but it seems to me to be a big step forward, and quite usable in some situations. It's not going to replace Tom Jones or Natalie Cole just yet, or The Sixteen for that matter... but does that mean we can't recognise and appreciate the progress made?

[James Perrett]

You may want to look back - and smile- as Herbie Hancock uses a vocoder so he can sing.

But he wasn't trying to make it sound like a real voice - it was an obviously effected voice and part of the sound.

And in any case a lot of modern auto tuned real vocals already sound unnatural.

Hence my last sentence - there are pop genres where an over-tuned vocal is acceptable and even expected. However, many people are thinking that these tools can be used in genres where vocals are expected to be more natural - which simply isn't the case at the moment.

There are many many many good singers around who could sing your songs far better if you just reach out to them.

[OneWorld]

Surely anyone takes it for what it is, an artificial representation of a natural phenomena? I'm sure those that write this software don't set out to banish all singers from all the musical world. In writing the software, obviously comparisons are made with the real thing, what else could an artificial singer be compared to?

There are those amongst the music community that write a tune and want to hear it sung in tune etc they can't do it themselves, so an artificial alternative is now available and pragmatism inspires that choice to use the artificial option.

And it should be remembered, AI, especially in creative industries, is very much a nascent technology and a work in progress. When any new technology is introduced there are the naysayers, buy the scientists plod on to try and develop a product that lives up to its claims and offers acceptable and affordable options. I've not found one yet, won't stop me looking though, we all learn to adapt.

[ajay_m]

As I understand it, Herbie actually did use a vocoder because it meant he could sing, something he otherwise wasn't naturally gifted at. He wasn't specifically going for the effect. I may be misremembering this, but I do recall reading him saying something to that effect at the time.

[Sam Spoons]

I believe you are right , but it was never expected to make him sound like a 'real' singer so no comparison with a 'real' vocal required.

It seems like only yesterday we were discussing 'Autotune' and how it would destroy the music industry... Now we all use it, maybe sparingly, but as and when necessary and appropriate. I listened to that demo and didn't like it at all but Mrs S has a couple of Michael Buble CDs and I don't like them either. They are, however very successful...