Synthesizer V, Felicia & Riku

[N i g e l]

2 New data banks available, Felicia & Riku

https://forum.dreamtonics.com/t/news-ou ... d-riku/293

“Synthesizer V AI Felicia (native language: English) exhibits a powerfully classic, richly operatic yet modern and steadily versatile female voice, injecting a ray of vocal sunshine into your productions and offering a wide selection of timbres - Airy, Bright, Musical, Operatic, Powerful.”

She would probably do well on ITVs “The Voice” 

https://www.youtube.com/watch?v=UiPNKFROPLQ

“Synthesizer V AI Riku (native language: Japanese) brings about a sensitive, crisp yet brittle-sounding male tenor voice and comes with a versatile selection of timbres, with Soft, Warm, and Powerful vocal modes.”

https://www.youtube.com/watch?v=vkYmN9s-0UY

[RichardT]

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

Sometimes Felicia sounds quite realistic, and sometimes she’s still in the uncanny valley.

[N i g e l]

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

Sometimes Felicia sounds quite realistic, and sometimes she’s still in the uncanny valley.

If you have the idea prepared beforehand, you can just tap in the notes over MIDI (or draw them in graphically) and add the lyrics in bulk. Synth V should sort out everything and get you 90% of the way there.

There will be some correction work due to phonetics and timing but most of the extra work will be the expression & emotion, Synth V doesnt know if you want a breathy whisper or powerfull crescendo. Thats the bit that takes the time.

[OneWorld]

That is so convincing. Of course we cannot perceive AI supplanting the human singer but for boxroom beat makers who can’t hold a note but cannot justify paying a pro-vocalist, though I do not in anyway think a vocalist should undersell themselves I hasten to add, this AI singer provides for the opportunity to at least have an ‘vocalist’ that can give the opportunity to hear our compositions sung in tune.

[N i g e l]

Time has passed & so more user demos of Felicia are available.

I like this one

[Synthesizer V Cover] What's Up? (4 Non Blonds) [Felicia]

https://www.youtube.com/watch?v=jyBcDtdwKkk

This one, Hmmmm, Im not sure if its because the singer is not suiting the song or because I liked the original Jessie J rendition. Whatever, if you listen to the original you can hear the differences.

[Synthesizer V Cover] Price Tag [Felicia]

https://www.youtube.com/watch?v=cBogJWSe550

both by uTuber " yossheee "

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

well you asked !... so heres is a half hour chat with the composer of the Felicia demo for the product release.

How we made Felicia's demo song | Synthesizer V Project Deep Dive

https://www.youtube.com/watch?v=DearfTudnmA

[N i g e l]

Synthesizer V Studio 2: Exclusive Preview

https://www.youtube.com/watch?v=MU_7vgvRbkY

Miguel & Kanru NAMM it up, demoing & discussing the work_in_progress Ver2

01:50

bigger, better, faster

50:00

When ? cost? discount ? - no comment, its work in progress

[sc1460]

Thanks for posting, such admiration for these guys and their skills to do this. I have version 1 and use it for backing vocals when singers aren’t available. They are 90% there!
Intrigued by sonarworks vocal AI, which might be less hard work for use on “traditional” genres, tried the demo, again for BVs, it’s astonishing. It’s only a matter of time now before both these plugins achieve 100% realism.
We are headed for a brave new world!
Cheers
H

[N i g e l]

V2 is scheduled for release on March 21

Available On March 21:
Synthesizer V Studio 2 Pro
We are thrilled to announce the forthcoming release of Synthesizer V Studio 2 Pro, a major update to our industry-standard song and vocal production software.

First unveiled at NAMM 2025, Synthesizer V Studio 2 Pro now offers new and improved voices, a more human-level of naturalness, lightning-fast processing speeds, plus plenty of innovative new functionality for all types of demo and music producers, songwriters and vocalists.

Synthesizer V Studio Pro will be available to purchase from March 21st. In the meantime, you can find out more and pre-order the software, along with all version 2 voices, bundles and upgrades.

[ajay_m]

It is a remarkable product at a surprisingly reasonable price point. You do have to put some effort into adding pitch and timing information to make the result more like a real singer, but the end result can be very realistic, as long as you don't ask your virtual singer to work beyond their natural range.
Sometimes you have to faff around a bit with lyrics to get consonants to have the right amount of presence and not get 'swallowed' but there's a lot of fine-level control available. I did find my Streamdeck useful to set up some key bindings so that it was a bit easier working inside the edit window and controlling the transport without a lot of mouse clicking, and I wish the dark colour scheme was configurable because I'm old-school and I like a light background and dark text, but it would be unfair to single Synth V out for criticism here given how many other products (looking at you, NI) aren't themeable or properly scaleable either.

As was mentioned in the original SOS review, I'd purchase Solaria along with the base product to get started as she is a very versatile singer, and then think about adding further vocalists once you've got your head round the product.

[The Elf]

I did find my Streamdeck useful to set up some key bindings...

As a user of both these products, I'd be interested to know some details!

[ajay_m]

I did find my Streamdeck useful to set up some key bindings...

As a user of both these products, I'd be interested to know some details!

Annoyingly, StreamDeck profile files are binary, otherwise I'd just paste it in here. What I did was (using the 15 key version) I mapped (for Reaper) the transport controls <<, >>, stop, play, pause, record to standard MIDI MMC messages, using the MIDI plugin, and Reaper responds to these directly.
[ these are just short sysex messages defined by the MMC standard ]

Then I mapped the internal Synth V functions for record, <<, >>, stop, play and pause to keyboard bindings (ctrl+shift+', ctrl+',shift+',ctrl+alt+',alt+', ctrl+1 respectively)

This makes it a bit less of a faff to manipulate either Reaper's main transport or the transport inside Synth V's main edit page.

[ajay_m]

oh, and looks like the upgrade cost for existing users will be USD39, which is rather reasonable. Plus you get to choose one of the new/upgraded vocalists as part of that cost, I believe.

[N i g e l]

that seems reasonable

...also on March 21st, there will be a new voice released

"Liam"

https://www.youtube.com/watch?v=MBI8AM19rU8

[The Elf]

Thank you for clarifying your approach.

It's surprising how even simple, but frequently used, key combinations allocated to Stream Deck can significantly help to speed up work. My Cubase profile makes me seem extremely slick!

[N i g e l]

V2 is out at the weekend

some updated V2 voices..

https://www.youtube.com/@DreamtonicsCoLtd/videos

seems there will be an initial deal, time limited, update the synth, update your voices & get a new voice, with change from the price of one old voice.

[N i g e l]

Phew, its a bit of an ordeal ! update the enigine 1st then do every thing else from inside.

V2 upgrade prices : for 1 month only..
$39 for the new V2 synth engine/editor
$19 per [dreamtronics] voice upgrade [less for bulk]

Brits, deduct for $ to £
add 20% for VAT

Solaria, £16, is late to the party it seems, being a 3rd party voice from ecliptical sounds

[N i g e l]

Solaria II now available. As update or fresh purchase.

[N i g e l]

PAH !

I was well on target to produce an awesome demo circa xmas 2027
but it seems ive been pipped at the post.

https://www.youtube.com/watch?v=Xv8RFB3bPKI

Solaria 2 mogen heap cover headlock

et also..

https://www.youtube.com/watch?v=dVnqfOjQMFc

yeah what ever !

[RichardT]

Impressive!

[OneWorld]

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

Sometimes Felicia sounds quite realistic, and sometimes she’s still in the uncanny valley.

Think of all the work Mother Nature put into creating a voice engine, that can handle emotion, expressiverness, timbre, dynamics, clarity, note perfect reproduction of the lyrics, and that voice engine has at its disposal a vast lexicon on tap, in real time and then articulate the result in a purpose made voice track that can work over several octaves, go from soft to loud effortlessy, from whimsical to profound, in perfect tune and timing.

Replicate all that and you'll appreciate that yes, there's a bit of effort required to produce convincing results.

If yo come across some software that can reproduce the voice even half as good as the organic representation, please let me know, because all I have come across so far good, but not good enough and I come to the conclusion that it would be easier, and more effective to just roll up the sleeves, settle down for long sessions of hard graft, and to learn to sing.

[OneWorld]

As a slight aside, if I may. Yesterday it was reported in the news that at some university in the US, an AI model had passe the Turing Test - a human conversed with a computer and it was thought by the majority of human speakers, had thought they were in conversation with another human.

That being said, one has to ask "What sort of dialogue was there exactly?"

And did they do a reverse test, did the ChatGPT invoke a conversation with both a human correspondent, and an artificial one?

[RichardT]

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

Sometimes Felicia sounds quite realistic, and sometimes she’s still in the uncanny valley.

Think of all the work Mother Nature put into creating a voice engine, that can handle emotion, expressiverness, timbre, dynamics, clarity, note perfect reproduction of the lyrics, and that voice engine has at its disposal a vast lexicon on tap, in real time and then articulate the result in a purpose made voice track that can work over several octaves, go from soft to loud effortlessy, from whimsical to profound, in perfect tune and timing.

Replicate all that and you'll appreciate that yes, there's a bit of effort required to produce convincing results.

If yo come across some software that can reproduce the voice even half as good as the organic representation, please let me know, because all I have come across so far good, but not good enough and I come to the conclusion that it would be easier, and more effective to just roll up the sleeves, settle down for long sessions of hard graft, and to learn to sing.

None so far - it in a few years I think things will be different. Personally, I am happy to use some AI voices now in mock-ups (Cantamus).

Using AI as a ‘performer’ feels very different from using AI to compose music, which doesn’t interest me at all.

[OneWorld]

Does Synthesizer V generate sounds like that just from notes and lyrics? Or is manual work need to polish the results?

Sometimes Felicia sounds quite realistic, and sometimes she’s still in the uncanny valley.

Think of all the work Mother Nature put into creating a voice engine, that can handle emotion, expressiverness, timbre, dynamics, clarity, note perfect reproduction of the lyrics, and that voice engine has at its disposal a vast lexicon on tap, in real time and then articulate the result in a purpose made voice track that can work over several octaves, go from soft to loud effortlessy, from whimsical to profound, in perfect tune and timing.

Replicate all that and you'll appreciate that yes, there's a bit of effort required to produce convincing results.

If yo come across some software that can reproduce the voice even half as good as the organic representation, please let me know, because all I have come across so far good, but not good enough and I come to the conclusion that it would be easier, and more effective to just roll up the sleeves, settle down for long sessions of hard graft, and to learn to sing.

None so far - it in a few years I think things will be different. Personally, I am happy to use some AI voices now in mock-ups (Cantamus).

Using AI as a ‘performer’ feels very different from using AI to compose music, which doesn’t interest me at all.

It hadn't occurred to use AI as a composer, I am not sure why anyone would bother trying to anyway, what would be the point?

[RichardT]

Lots of people will, I expect. But I agree, to me there's no value in music composed by AI.

The same with images generated by AI - I expect by now the tools can generate a mock Leonardo da Vinci painting pretty much perfectly, but for me there is no point or worth in doing that.

[N i g e l]

Lots of people will, I expect. But I agree, to me there's no value in music composed by AI.

but there is value in music composed with AI?
It's just a tool I use.

Learn to sing ? One of the reasons for synthesizers, is to be able to emulate other instruments using your interface of choice. If I want a harp sound, I dont want to have to learn the harp.

Another reason for synthesizers is to go beyond natural instruments.

(Thinking about the blue creature opera in 5th Element film, or adding a toggle switch to male VST voices "restrict testosterone before puberty"