Project_Beta: A multi-expression commercial voice for UTAU

Which bank would you like to make into the final? (green expression)

  • Growl

    Votes: 3 50.0%
  • Shout

    Votes: 0 0.0%
  • Whisper

    Votes: 2 33.3%
  • Fry (experimental)

    Votes: 1 16.7%

  • Total voters
    6
  • Poll closed .

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Last Update: 01/24/2016

Beta is a commercial voicebank for the Utau software.

4MHV7pj.png

temporary logo, subject to revision in the future

Beta's voice bundle is intended to be high quality and intended for producers and music creation. He is recorded on a private reclist created by Acoustic and proofread by DystoP(ia)
These lists utilize X-Sampa as a standard for phonetic input and to make the transition from other software easier.



Our Team:
Gamma: Project Overseer, Reclisting, Voice Provider.
DystoP: Reclisting, Testing, Feedback.

bDmByqj.png


Due to the size of Beta's voicebank he will only be sold as a physical package. He will be available as a DVD and as a USB drive.
Documentation and a Quick Start Guide will be provided with the product as a PDF and translated to several languages.
Beta comes in 48kHz samples and utilizes Kanru Hua's "Moresampler" back-end.
Those who want to use Beta with other engines have the option to install a modified version of his voice at time of installation.

Contact Email: Gamma@dynamivox.xom
Website: Under Construction
Blog: blog.dynamivox.com
KCaJbIY.png
PIIzmfa.png
c7DcJEb.png
skz3G9C.png
fK1yr1N.png
D8Ag1ju.png
PfSQxdu.png

None of these icons currently do anything, Stay tuned!

News/Announcements:
Beta is now going to have an 8 pitch VCV+CVVC Japanese bank included. Thank you for your interest. A new reclist has been made to allow for more natural recording.
Extra sounds have been added to both his English and Japanese to allow for some additional language capabilities. This includes Spanish.

PROJECT GOALS:
These are non-critical project goals, whether these are obtained or not shouldn't have too drastic of an impact on it's final outcome.
  • Bundle shareware UTAU as an optional package.
  • Expand Beta to a 15 sub-expression bundle.

We are recruiting team members!
While we are still quite early in development, we are still looking for future members for this project!
All team members will be paid for their service and need to be decently proficient in the aspect they are applying for.
Thank you for your interest! The types of skills we are looking for are listed below.
We will contact you later at a later point of the project regarding your application.

Configuration-
A big part of creating Beta is the Voice Configuration. As a very large bank this is the job that requires the most individuals. Upon acceptance you will be provided with documentation regarding the project configuration standard, these are required to ensure the voicebank(s) are configured uniformly and consistently.
You must be proficient in VCV and/or CVVC and may be teamed up directly with another team member to fulfill different parts of the assigned configuration task.
You would be given one month to finish an assigned task.

Demo Tracks-
One a large portion of Beta's voicebank(s) are completed, songs demoing his voice will be needed- Due to the nature of this task we may not take direct applications.
Producers will be sent a physical copy of a "producer demo" version of the voicebank when a workable version of the bank is available and will be sent the final version for free when the project is complete.

Beta Testers-
Sitting along with those tasked with making voice demos, Beta testers will need to test the voicebank for faults that affect it's quality. Upon fixing we will distribute a small patch to download and incorporate it into the final voice.
This is a dual-task and is also asked of the producers demoing the bank to also test and report issues.
To prevent a flood of Beta Testing requests, applications for this are not accepted.

Documentation Translation-
The Guides and Documents included with Beta are initially only available in English and will need to be translated to various languages (primarily Spanish, Japanese, and French)

When applying for ANY of these please provide detailed examples of your previous experience in that task! You will be recorded as potential team members and will be contacted further at a later date.


How We Started:
Many events have taken place which resulted in the project's current state, for the time being a brief summary shall be left here.
This project started as a result of wanting to provide a high quality voicebank and was originally going to be provided free. After much planning it was realized that the result we were going for couldn't be provided free-of-charge and we shifted our plans to offer Beta as a sold product.
Since then much effort has been put in to improve our plans and expand the bank's capabilities.

Discussion is greatly appreciated, wel''l try to answer any questions you have and update our post!
Suggestions and ideas welcome.
 
Last edited:

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Beta will be provided as a collection of several voicebanks with their own capabilities, the primary one being a general voice with additional expressions made available to expand upon it.
He will include two series of banks, one English and one Japanese, both which have extras added to increase the multilingual ability of both of these.




The following Expressions will be included with Beta:
Core: Standard voice
Core +
Core -

Cyan: Gentle Expression
Cyan +
Cyan -

Magenta: Strong/Solid Expression
Magenta +
Magenta -

Yellow: Bright Expression
Yellow +
Yellow -

Black: Dark Expression
K +
K -

Green: Growl (Japanese only)

Each of these banks will be 7 pitch (the Japanese being 8) and have a similar pitch.
the +/- denote close variations of base bank and are still completely separate banks of their own. + being more forward and stronger and - being weaker and raspier while still fitting with the base bank. These can be used to obtain the specific tone you would like for a song or to creatively transition between the expression sets in an emotional way.

If you have any questions about the project/voice, feel free to ask.
 
Last edited:
  • Like
Reactions: Kiyoteru

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
A cyan Japanese example will be previewed in a few days, stay tuned!

Also a version of the thread is now available of VO.
 
Last edited:

HulderBulder

Retired User
Retired User
Defender of Defoko
Can I ask how many spaces are between the pitches? Considering its a 9 pitch it seems like alot of work, and it may or may not be usefull depending on the space between them and tone.
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Can I ask how many spaces are between the pitches? Considering its a 9 pitch it seems like alot of work, and it may or may not be usefull depending on the space between them and tone.
The pitches are spaced based on tone of the voice.
Near the bottom of the range each pitch is only 4 semitones apart maximum due to how utau handles lower notes, in higher ranges the pitch sets are spaced a bit more.
if it wasn't useful then we would of excluded it and saved the disk space.
 
Last edited:

HulderBulder

Retired User
Retired User
Defender of Defoko
The pitches are spaced based on tone of the voice.
Near the bottom of the range each pitch is only 4 semitones apart maximum due to how utau handles lower notes, in higher ranges the pitch sets are spaced a bit more.
if it wasn't useful then we would of excluded it and saved the disk space.
ah that explains it, thanks for clarifying.
 

Zoku

making doper vocaloid music than the rest
Defender of Defoko
I kinda like the CMYK aspect to it--

The current sample sounds really nice and clear, soothing even. If I had money, I would buy it. Alas, I am a poor teenager. I look forward to seeing the finished product!
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
After much thought, Japanese will no longer be supported in Beta's voicebank.

Whether a separate voicebank is to be included is undecided.
 

Zoku

making doper vocaloid music than the rest
Defender of Defoko
I was looking forward to Japanese VCV... Was there any specific reason for canceling it?
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Update: Beta will now include a TTS voice free with his voicebanks, this voice will be built on the Festival speech engine.
This will allow Beta to speak (English only) if you so choose.
-
While we will not be officially supporting it, singing synthesis with Festival/Lyricos/Flinger using his TTS voice is possible for Linux and Mac users.
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Announcement:
We are currently encountering issues with the project at this time.
Due to the size of Beta's reclist we are having problems with UTAU's oto.ini size restrictions.

With this in mind, we are in the process of reworking this project to recover configuration space without affecting voicebank quality.

New updates regarding this will be posted shortly.
 
Last edited:
  • Like
Reactions: Kiyoteru

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Announcement:
A test for "Beta for Festival" will be developed while UTAU bank development is looked over.

In the meantime there may be a lack of updates as voice creation for Festival takes some time and provides no provisions for testing and usage until it is finished.
This voice is intended for speech but will fully support Festival's singing synthesis.

While this voice will be included free with his UTAU voice products; if you choose to purchase it separately then a portion will be donated to Festival (it's team) and all tools involved in Beta's Festival voice creation.
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Updates:
-Spanish "ll" (voiced palatal fricative variant) added.
-Alveolar lateral flap added to list.
-M, N, and L are now treated as vowels (replaces n/m merger)
-Formatting fixes
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
I must apologize for the lack of audio updates, I've been sick and I have been avoiding recording.
So I'm using this time to do housekeeping.
 
  • Like
Reactions: Kiyoteru

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Many problems have arisen that make this project increasingly difficult even in it's reworked state, for this reason this project may see no public updates for up to 6 months.
Most of the work from earlier in this project must be scrapped and this holds back Beta's expected release drastically.

Development will still be done behind the scenes when possible but updates will not be posted publicly. I will try to answer questions when possible.
 
Last edited:

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
Updates made to original Post.

Time for a slight vote/discussion.
The green field in the chart on the first post is the estimated range of a possibly included voicebank.
You can choose between a fully functional growl, shouting (Angry/Rage), gentle whisper, or vocal fry voicebank.
  • A growl VB would use tn_fnds to loop the samples and has worked very well in testing. Fully handling pitch changes.
    A growl bank would have a limited range of no more than C4.
  • A shouting bank would be literally shouting, this would have some raspy/harsh tendencies as it's essentially angry shouting.
  • a Gentle Whisper bank would be borderline un-voiced but still fully supports pitch shifting. In this case the bank would actually include a range of E2 - G4
  • A vocal fry bank would be functional but quite limited, this is still largely untested.
Four options. While the potential of these banks has been considered only one can actually make it into the final bank.
 

oteto

Tetoholic
Defender of Defoko
vocal fry, i think is pretty rare to find? the angry shouting and whisper banks sound like they would be good additions as well! maybe soon we can hear some samples of these!
 

na4a4a

Outwardly Opinionated and Harshly Critical
Supporter
Defender of Defoko
Thread starter
In testing generally everything is tested in nonsense tones and random sounds.

If that's what you'd like to hear then sure, I'll post some examples.