Synthesizer V - Singing voice synth for Win/Linux/Mac (Free & Payware)

[ZincT]

Thanks, useful information. Cheers koukouba

[Spitfire31]

- Can be used as VST plugin inside your DAW

And as an AU plugin in a Mac environment. Good news for Logic aficionadoes!

[Pelicanomicon]

Synthesizer V Studio 1.5.0 incoming at 1800 JST today with further improvements to AI, including an amazing update: multilingual capability for AI voicebanks! Chinese voices singing in Japanese, Japanese voices singing in English, dogs and cats living together...Mass hysteria!

---

https://twitter.com/ahsoft/status/1473852484439666689 (Japanese)

Press Release: Synthesizer V Studio Pro supports multilingual singing with the latest update of Synthesizer V, releasing today. Sing in English and Chinese with Japanese voice databases, and Japanese and Chinese with English voice databases.

---

https://twitter.com/uchuuzentai/status/ ... 6348354564

Quadimension [a Chinese SynthV vb developer] have announced that today at 5PM (China time), all 6 of their R2 [Standard] voices will be upgraded to AI and will be able to support Japanese and English in addition to Chinese.

ETA woops, @uchuuzentai issued a correction:

Update: Today at 5PM (China time) will be the update for AI databases to support other languages. Since they need to train the Standard MEDIUM5 voices for AI, they need to be repackaged and this will take some time.

---

Example of Eleanor AI singing in Japanese from this DTM Station article, "Indistinguishable From Human: The Overwhelming Power of Synthesizer V, Now With Multilingual Singing Added"
https://www.dtmstation.com/archives/54473.html (Japanese)

https://www.youtube.com/watch?v=Cc1M51XMFgw
fwiw Eleanor does have a noticeable accent, but according to the article it seems to be a deliberate touch from Kanru Hua and Dreamtonics.

Kanru Hua:

完全に日本人の発音になると元のキャラクターと似てないと言われる可能性も考えて、微妙な発音やアクセントが残してあります。そこがちょうど良いバランスではないかと、考えました。ただ、エレノア・フォルテで日本語を歌わせると、母音のaは口の開き方が少し大きくなるとは思います。それは英語の発音のaと日本語の発音のaに違いがあるからです。ただ私が聴く限りでは完全な日本人発音のように感じますが、それは、私が外国人だからですかね？

I thought about the possibility that if I were to completely change the pronunciation to Japanese, people might say that it doesn't resemble the original character, so I left in some subtle pronunciations and accents. I thought that would be the right balance. However, I think that if you make Eleanor Forte sing Japanese, the vowel "a" will have a slightly larger mouth opening. This is because there is a difference between the English pronunciation of "a" and the Japanese pronunciation of "a". But as far as I can tell, it sounds like a completely Japanese pronunciation, but that's because I'm a foreigner, isn't it? -- Translated with www.DeepL.com/Translator (free version)

DTM's examples of Eng-to-Jpn and Chn-to-Jpn are "easier" to simulate. I am DYING to hear how this works for Japanese to English, the phonology is so different.

More importantly, if this works well there will be more voice types for consumers. Quadimension has one of the only two male voicebanks commercially available right now, "Muxin".

It's a Christmas miracle!

[Pelicanomicon]

Updated without any chugging and did a simple test of Saki AI, my only Japanese SV bank.

...Oh no, at first glance this thing sounds pretty good. Better than any non-native English Vocaloid I own, and better than many of the native English Vocaloids I own. Gonna experiment more later, got holiday stuff to do.

[ZincT]

Thanks for the heads-up Pelicanomicon. I have now updated and hope to test it out later.

[Pelicanomicon]

Official Dreamtonics demonstration of the feature "Cross-Lingual Singing Synthesis" w/ teaser for male English voice coming early 2022
https://www.youtube.com/watch?v=De4cnZkl5pQ
Crazy. CRAZY. What is Yamaha going to do for VOCALOID6? all that money i gave them aaaaarrrrrrrrghhh

Press release
https://dreamtonics.com/en/cross-lingua ... izer-v-ai/

This is supposed to be a demonstration with Quadimension Cangqiong vb. I always liked the tone of this one, hope AniCute carries it
https://www.bilibili.com/video/BV1134y167fc

[Spitfire31]

This is supposed to be a demonstration with Quadimension Cangqiong vb. I always liked the tone of this one, hope AniCute carries it
https://www.bilibili.com/video/BV1134y167fc

Cangqiong sounds stunning and with an 'emotional depth' (for the lack of a better word). She would be a great asset in SV!

[Pelicanomicon]

More Cangqiong AI English testing from Creuzer, the same tuner from this post
https://twitter.com/Creuzer1/status/1474262759676149767

Could we please get a male AI that's the tonal equivalent of this?

According to the SynthV wiki, the Quadimension AI voices are being trained by their Standard (hybrid concatenative vb) recordings. The possibilities for older/discontinued Vocaloids...

[Spitfire31]

More Cangqiong AI English testing from Creuzer, the same tuner from this post
https://twitter.com/Creuzer1/status/1474262759676149767

Amazing! I think I'm in love…

For once, a grown-up voice, cut'n'dried for rock and country!

[Spitfire31]

More Cangqiong AI English testing from Creuzer, the same tuner from this post
https://twitter.com/Creuzer1/status/1474262759676149767

Could we please get a male AI that's the tonal equivalent of this?

Yes please. Someone like Brett Sparks from The Handsome Family? You'll recognise the song, I think.
https://www.youtube.com/watch?v=p4zluA60hjs

[Pelicanomicon]

For once, a grown-up voice, cut'n'dried for rock and country!

It is a great fit for country, a genre that I've been forever wanting companies to make a voicebank for. Maybe this could start a renaissance of new voicetypes. I wonder how much a cinematic SATB set would cost to make?

Could we please get a male AI that's the tonal equivalent of this?

Yes please. Someone like Brett Sparks from The Handsome Family? You'll recognise the song, I think.

Funny, when I posted I was thinking "Does Clancy Brown sing?" and this is a perfect, real example of the sound I imagined. Thanks

[markmann]

Apparently version 1.5.0 with Cross-Lingual Singing Synthesis is now available. I just downloaded it after clicking on "update" in the standalone version.

I just bought this 2 weeks ago along with Eleanor AI and finally went to try it out when I saw this update. So I'm not familiar enough with the program to assess this new version. But if it actually makes all those other voices available in understandable English, that will really be great. This is one of the most exciting software advances for music in quite some time.

[Pelicanomicon]

Eclipsed Sounds will start taking pre-orders for SOLARIA on January 1, at a price of US $89.99. Here is a new a cappella demo showing off Synthesizer V's cross-lingual capability* and auto-pitch tuning for the voicebank:
https://www.youtube.com/watch?v=3TkWy6CTVJU
Eclipsed Sounds Twitter thread with more info
https://twitter.com/eclipsedsounds/stat ... 0339613700

*Cross-Lingual Singing is an exclusive feature of Synthesizer V Studio Pro and (at this time) commercial AI voices.

[Spitfire31]

Solaria will do until Cangqiong AI is released in the wild.

Will the cross-lingual technique allow, for instance, French, German, Danish, and so on versions?

[VitaminD]

I just found this thread despite it being around for years it seems. This product is rather hidden in plain sight. And it is a bit difficult for me to navigate the Japanese language website to understand what to download to get this work. But once I did, it is really interesting!!

The English still sounds a bit synthetic, but is really really nice compared to other tech I've heard in the past. Could fit well in certain genres currently and in others, where it might not, could at least provide an aural example for a human singer to base their singing off of.