Other DeepVocal

[Twillby]

(VNN Article summarizing the news in case this OP is too much of a mess I TRIED)

On June 7, Boxstar, the creator of Sharpkey, posted an announcement on Twiitter that his new project DeepVocal was ready for beta, invited people to submit voice introductions if interested in participating, and posted a brief video of a monopitch bank. He has also stated that it has been tested with Mandarin, English, and Japanese.




I also found a demo on Youtube

 

[Kona]

I found two other demos, including an English male!


I really like how DeepVocal sounds! I’m quite excited about it

 

[xuu]

Boxstar finally lives again it is time to celebrate

Anyways, this is what he's been working on for months and is the reason we saw no new Sharpkey releases in forever. DeepVocal is basically the new name for Sharpkey, it just has an updated engine. Actually, most SK VBs (bar Inari, they're working on that) work in DV. All the VBs that were being worked on for Sharpkey will be released for DeepVocal instead, as far as I'm aware, including Rabia (demo) and Chorical's Yu Sheng. The new vocals are Xia Yezi (Japanese), Mo Yu (Japanese) and Wen Xi (English). No clue if Yezi/Mo Yu/Wen Xi will recieve VBs in other languages right now. Hopefully will have something to show that I have been working on in the coming months too, so stay tuned~

 

[VocAddict]

In the English demo, there seems to be some double aliasing going on as we see both X-SAMPA and VCCV being used for the phonemes. That's great honestly because it allows people to transition easier without having to worry about learning a new phonetic system.

 

[xuu]

In the English demo, there seems to be some double aliasing going on as we see both X-SAMPA and VCCV being used for the phonemes. That's great honestly because it allows people to transition easier without having to worry about learning a new phonetic system.

I think it's actually just a quirk of the reclist used, but I'll have to ask the guy about it... Also, it seems to lack things like ai, ei, oi, instead favouring combining the individual vowels themselves. I think it works better thanks to how DeepVocal works, but he's considering adding to the list.

 

[DefiantKitsune]

Lacking dipthongs seems strange to me: they're basically mandatory for natural pronounciation in languages that have them, and unlike certain synths from Spanish/Japanese devs, Chinese actually has dipthongs.

Perhaps the software can recognize when a dipthong is needed?

 

[xuu]

Of course I'm unsure to how it fully works due to currently lacking the development tools, it seems to be the case that the software recognizes when one is needed. Though, adding them as actual aliases/recordings is possible.

 

[YunīC]

The new vocals are Xia Yezi (Japanese), Mo Yu (Japanese) and Wen Xi (English). No clue if Yezi/Mo Yu/Wen Xi will recieve VBs in other languages right now.

Since Xia Yezi is a Chinese synth for Niaoniao/UTAU I think she will probably Chinese & Japanese like Rila.
(Though I don't know for sure eighter)

 

[Exemplar]

 

[frankensalad]

yaaaaaaaaay

 

[xuu]

Hi everyone! I've been a bit quiet because of this. Like I said in an earlier post, I was hoping to have something to show in the coming months, and this was a part of it. A few days ago I recieved the DeepVocal ToolBox and after a very long night and a lot of work today, I have an initial demo to show. Introducing Lian Ye Yao:


A couple of people will know I've wanted to develop on Sharpkey/DeepVocal for well over a year now, and a while ago Kiki asked me to voice Lian Ye Yao, and to get him and his twin onto DeepVocal. Finally I can say that's a success. This is still early beta (marked as v0.2!) and only has one pitch with lots to optimize, but it works and I really wanted to show something. I'll try to answer anything you may have to ask that I can.

 

[Infoholic]

Hi everyone! I've been a bit quiet because of this. Like I said in an earlier post, I was hoping to have something to show in the coming months, and this was a part of it. A few days ago I recieved the DeepVocal ToolBox and after a very long night and a lot of work today, I have an initial demo to show. Introducing Lian Ye Yao:


A couple of people will know I've wanted to develop on Sharpkey/DeepVocal for well over a year now, and a while ago Kiki asked me to voice Lian Ye Yao, and to get him and his twin onto DeepVocal. Finally I can say that's a success. This is still early beta (marked as v0.2!) and only has one pitch with lots to optimize, but it works and I really wanted to show something. I'll try to answer anything you may have to ask that I can.

We stan

 

[kozet]

I'm excited to see what'll come of DeepVocal. I hope there'll be a Linux version available later on, though.

 

[kozet]

Apologies for double-posting, but DeepVocal has been released! However, the download link is a bit flaky at this moment.

 

[Mei-Saime]

I decided to make Yuzuko DeepVocal voicebanks and this is a little voice demo.

 

[Exemplar]

 

[Exemplar]

This is going to be interesting

 

[Mei-Saime]

Hello. As the main person in charge of Re:Voice, I'm making an announcement. This upcoming cover of "蜜月アン・ドゥ・トロワ" will be featuring Yuzuko's 1st demo with her 1 closed beta VB which is at 50%. The full cover will feature a simple PV and detail of the closed beta will be posted on Re:Voice's official Twitter and Weibo accounts on July 23rd. You can catch a sneak preview of the cover here on BiliBili or a special scheduled released on YouTube. Follow @/VoiceAzure on Twitter or [Sina Visitor System] on Weibo for details of the closed beta.

 

[Mei-Saime]

Re:Voice's first vocalist, Yuzuko is now released for closed beta.

 

[Infoholic]

New update for DeepVocal as well as the DeepVocalToolBox;

DeepVocal

DeepVocal Official Website,ディープボーカル 公式サイト,DeepVocal 官方网站

www.deep-vocal.com

Editor is in ver 1.1.0 now which improved synthesis
ToolBox is in ver 1.1.2 which fixes model building as well as general quality of product.