CeVIO shows off AI Singer’s development status. When will the software with GUI come out!

AIシンガー

Recently, AI singer “Kiritan” has been released and AI singer’s fever has been increasing in the area of voice synthesis software.

In November 2019, Techno-Speech launched an official account and started SNS promotion activities.

Official Twitter account of Techno Speech

Check out the recently posted information related to “CeVIO Pro (tentative)”.

This content is current as of March 12, 2020.
This text has been automatically translated using DeepL.
JPN→https://km4osm.com/cevio-pro-kari/

CeVIO Pro (tentative)

Techno-Speech has been developing HMM-based voice synthesis, but recently it seems to be moving towards commercialization of voice synthesis using AI technology (such as CNN).

This is also a software that sings like a human being if you give it a sheet of music. And it has a GUI! It allows the user to tweak the pitch and whatnot.

One composer said that even NEUTRINO, which does not have a graphical user interface (GUI), can be used in his work, so of course professionals would be very happy if it had a GUI.

It’s named “Pro”, but it probably shouldn’t be dedicated to professional use. (It’s also unlikely that the CeVIO Pro will be priced for the pros.)

Notable points (1) The familiar voice of Sasara

When I listen to the audio demo, I hear the familiar voice of Sasara.

If it’s this voice, “It’s not Sasara’s voice! I don’t think there are many people who would yell at me. I’m embarrassed if I’m wrong, because the Techno-Speech didn’t explicitly say “this is Sasara’s voice”…

There is more stability in his voice than what I heard in the demo at the end of ’18 or on the DTM station CD. It’s hard to get out of tune, and it doesn’t sound like there’s much of a blur in the volume.

DTMステーションのAIシンガーCDを聞いてみた|くろ州の合成音声備忘録
今日は、同人音楽イベントM3で私が買ってきたCDの感想をお送りいたします。【前置き】2018年末にテクノスピーチから「新しいAIシンガー技術ができたよ」という感じのリリースがあったんです。 『AI技術により超高精度な歌声合成を実現 ​~バーチャルシンガーの歌声は人と区別できない時代へ~』で、この技術を使ったCDが出ると...

As a person familiar with VOCALOID’s voice, it is very easy to listen.

Put differently, it sounds like Sasara’s voice, so even if you’re told this voice without any previous information, you won’t mistake it for a human voice.

Notable points (2) – VSTi

Surprisingly, it has perfectly implemented the key element for composers, “linkage with DAW“. It runs as a plugin (VSTi) on REAPER or Cubase.

A composer can make revisions as he writes a song. Isn’t Techno-Speech seriously trying to get professional users?

They’ve been saying for a long time that I’d like to work on the DAW linkage for a long time, and it seemed to be ready as of last year.

NeoCeVIO(仮)見てきた|くろ州の合成音声備忘録
 CeVIOエンジンを作ったテクノスピーチさんが2018年にディープラーニングで人っぽく歌える歌声合成を発表しましたよね。あれを見てきました。AI技術により超高精度な歌声合成を実現 ​~バーチャルシンガーの歌声は人と区別できない時代へ~【経緯】テクノスピーチNakamuraさん「最新の歌声合成のデモします」私「行かねば...

On the other hand, for amateur voice synthesis software users (especially those who specialize in covers) and DTM beginners, VSTi-type voice synthesis is a high hurdle in some respects. I predict that there will be some demand for the standalone version.

Notable points (3) – Mac support

Dear Mac users! It looks like CeVIO will be available on the Mac soon. This is a great opportunity to get new customers.

There are some Linux DAWs that support VST, so it might work on Linux.

There are some DAWs for Android smartphones that support VST, but running CeVIO on an Android smartphone seems to be difficult.

Notable points (4) – Parameters

The following parameters can be seen in the image that is only momentarily reflected in the demo video.

・NOMAL
・PITCH
・VOLUME
・VibratoAmpliftude
・VibrateFrequency
・Duration

The picture quality is so low that I don’t know exactly what it says.

If I’m right, there seems to be a CS6 equivalent parameter at the moment.

If you are a CeVIO user, you can use your past experience and techniques as they are. For those who are not comfortable with the current CeVIO, it’s still going to be tough.

Notable points (5) – What about the talk?

There is currently no information about CeVIO Pro Talk (tentative) at all. I didn’t find any Talk-related parts in the GUI on the demo.

Techno-Speech shouldn’t have released an audio demo of Text-to-Speech to the public since the end of ’18.

Talk should be selling well, and I don’t think it will disappear, but I can only imagine that CeVIO Pro (tentative)’s development policy seems to lean more towards singing voice synthesis (VST support is not necessary if the focus is on talk), and maybe it will be sold in the same way as CeVIO Talk Pro (tentative) as a separate software…

AI singers who are easy to tinker with can be the most powerful.

Translated with www.DeepL.com/Translator (free version)

コメント

タイトルとURLをコピーしました