UTAU "Resampler" Directory

[Lightzaver]

VS4U is like BKH01 or any of the EFB resamplers, it just sounds more realistic and handles pitch bends better.

yea I don't like the new version of Vocalshifter

 

[Lightzaver]

Not sure if the main post includes the 2013 updates of some of the resamplers, but, I found a website with updates to some of them. But only WARP and EFB_GT- Adding flag support to WARP and i'm not sure of the changes to EFB-GT.

You can find them both, here- http://custom-made.seesaa.net/category/13425893-1.html

//Also, keep the old versions. I suggest you keep both as the new versions seem to amplify the slightest bit of noise in a voicebank. Even supposedly clear voicebanks seem to become noisy with the latest updates.

 

[shinami]

Ooooh, sweet, flag support! If only VS4U would get that now~

The changes to EFB-GT are now AIFF file support (MAC files I think?), improved STP working I guess, and use of the flags P and t.
WARP's changes are a similar STP fix and flags P t b B g H K (which does the same as fresamp K) and the option to have no format filter.

and yeah sorry about that i just don't really update this thread ever anymore ;;
Never did get EFB-PT or whatever to work...

 

[Lightzaver]

yea, they all work for me (?) apart from WORLD. and I have no idea why ;~;

The EFB's are pretty good, especially PT because its great with soft banks

 

[kimchi-tan]

I read the readme for VS4U and has it's own flags, apparently:

• T: t flag of 1cent unit (-2400 to 2400, base value 0) ~ so this is like the extreme version of the t flag
• I: Brightness (0-100, base value 50) ~ is this equivalent to the Vocaloid Brightness property? I don't own any Vocaloid so I'm not knowledgable about that property but increasing it makes vocals sound more lively
• A: Kurianesu (0-100, base value 50) ~ Similar to bkh01's A flags. Creates noise on some notes so best to not raise it high. It makes the vocals more powerful. It helped undistort powerful vocals in my case [size=-2](since I used Ted when I tested it, use it for powerful genderbent voices)[/size]
• M: speech synthesis algorithm (0-2, base value 0) - You may be improved if the noise is generated, by specifying the M2 or M1. ~ it improves the noise removal? I use this flag whenever I use this resampler. Seems great.

It also can be affected by:

• Intensity
• Modulation
• BRE
• No Format Filter
• Constant Velocity
• g, t, H, P flags

Also, it creates it's own FRQ files which will be created in a folder of the voice library. [size=-2]Although I can't find them OTL[/size]

It's one of those resamplers that takes pitchbends realistically and it's also a fast resampler, about the default resampler's speed.

I'm using it for some cover songs so I'll update this post if I find anything more. The latest version is 1.1.

 

[Umbreon126]

[retracted]

 

[Zoku]

Can someone post a sample for VS4U? I am interested in it and at the moment can't get it because of some PC problems... Thanks! (Comparison to resampler and bkh01 would be cool too, if that's ok.)

 

[Lightzaver]

I can do so, I'd also like to say that the flags introduced are very minimal, with little effect.

[soundcloud]https://soundcloud.com/zappierchris/resampler-test-bkh01-resampler[/soundcloud]

In this order: BKH01 -> Resampler -> VS4U

Using Aino Hate ( 愛野ハテ ) Ver 2. She is a very raspy, light voice. Please note I used no flags apart from breath and some constant velocity.

Even though BKH01 sounds really bad i'd probably add some flags like O75V50 e.t.c

 

[kimchi-tan]

I've been using it and it's really good if used right https://clyp.it/r0rk4s4x

I used the flags (for the rendering options) g+30(because it's Ted)H10Y30I80M2 and I50A100 for the powerful notes (sakebi).

 

[Zoku]

I currently use BKH01, and VS4U has so much difference! I like VS4U more than bkh01 now, it gives a more kinda-realistic feel to my UTAU and makes it flow better!

 

[kimchi-tan]

I can do so, I'd also like to say that the flags introduced are very minimal, with little effect.

[soundcloud]https://soundcloud.com/zappierchris/resampler-test-bkh01-resampler[/soundcloud]

In this order: BKH01 -> Resampler -> VS4U

Using Aino Hate ( 愛野ハテ ) Ver 2. She is a very raspy, light voice. Please note I used no flags apart from breath and some constant velocity.

Even though BKH01 sounds really bad i'd probably add some flags like O75V50 e.t.c

Hate has a really raspy voice like you said (I have her too ^^) and with Tokyo Teddy Bear being a powerful song, fresamp really is the top choice for the resampler (with the noise removal and vocal strengthening it provides).

I like the VS4U sample you provided the best but that's just my opinion.

 

[Lightzaver]

I can do so, I'd also like to say that the flags introduced are very minimal, with little effect.

[soundcloud]https://soundcloud.com/zappierchris/resampler-test-bkh01-resampler[/soundcloud]

In this order: BKH01 -> Resampler -> VS4U

Using Aino Hate ( 愛野ハテ ) Ver 2. She is a very raspy, light voice. Please note I used no flags apart from breath and some constant velocity.

Even though BKH01 sounds really bad i'd probably add some flags like O75V50 e.t.c

Hate has a really raspy voice like you said (I have her too ^^) and with Tokyo Teddy Bear being a powerful song, fresamp really is the top choice for the resampler (with the noise removal and vocal strengthening it provides).

I like the VS4U sample you provided the best but that's just my opinion.

I'm also starting to like VS4U, mainly because of the way the flags work. Although, I used no flags ;u;

 

[kimchi-tan]

Hey guys, I tried EFB-GW and it's REALLY screwed up [soundcloud][/soundcloud]

When I first download EFB-GW with the fftw files, it didn't work. But then I found this EFB-GW download at 4shared (I'm not sure what version it is) that came with the libfftw3-3 file, it worked... but it was screwed up. The notes I selected were Teto's whisper VCV and were supposed to be "ka ga~mi~ no~" but ended up being "ka ga~ sa mi~ mu ri no hi"

I later tried the resampler patch on it but it then it didn't work. I copied the libfftw3-3 to the resampler folder then it worked again but it had the same effect again Here's the sample with EFB-PB (what it's supposed to sound like) and EFB-GW (how it sounds screwed up in my computer): https://clyp.it/jai0murx

Another thing: EFB-PB takes longer than fresamp to render for me.

 

[Lightzaver]

EFB-PB is completely broken for me. For each note it renders, it stops responding and has to be restarted, so it's really annoying for 600+ note songs.

Anyway, WORLD resampler has moved.

Spoiler

WORLD resampler was last updated July 26, 2013. The website does not list changes, and the download it includes the source files, so you have to compile it yourself.

The FAQ also says something about a "Matlab" version of the WORLD resampler, but the author doesn't want to distribute it. It apparently only has some minor differences.

I've spent like an hour trying to compile/build it and the test.exe it outputs doesn't work :/ Oh well.

//Edit; if you can compile the file, you run it through cmd. So you can make it work for single files, but it doesn't work for whole songs, or in utau itself.

VS4U has been updated (2014/01/18) The M flag has changed.

Spoiler

Changelog:

ver.1.2beta1 2014/01/18
To change the specifications of the M flag.

ver.1.11 2014/01/02
Fixed formant correction algorithm. (VocalShifter ver.2.2beta2 equivalent)

The M flag can be used to improve quality as notes get higher.

M:音声合成アルゴリズム(0～1,基本値0)
1オクターブ以上ピッチを上げる場合、M1を指定すると音質の劣化を抑制できることがあります。

 

[irei1as]

There's a new "kind of" resampler called utaugrowl.
It's not exactly a resampler. It's more a wrapper of other resamplers that adds a growling flag (w0-100).
A description and the ux.getuploader.com download is in this NND video:
http://www.nicovideo.jp/watch/sm22646234

kimchi-tan explains its use in http://utaforum.net/thread-new-growling-resampler
but let me re-type it here.

How to use:
->Unzip utaugrowl.exe into your resampler directory (usually where utau.exe is).
->Rename utaugrowl.exe into @name_ug.exe. @name is the resampler you want to use as base (and it's in the same folder), for example: resampler_ug.exe, freesamp_ug.exe or oldresampler_ug.exe
->Use that *_ug.exe as resampler for the project properties and add the w flag for the notes you want "growling".

As bad news, it can't work with resamplers that has been accelerated with nmasao1's resampler patcher. We have to use original unmodified resamplers.

 

[Iscabird]

^ I would like to correct that second step because it will not actually work otherwise:

Do not rename utaugrowl at all. What you want to do is make a copy of an engine you already have (be it regular resampler or fresamp or even tn_fnds if you want, whatever engine you want), put it in the main directory with utaugrowl and add _ug to the copy's name. Then when you want to make the growling noise, set utaugrowl.exe as Tool 2, not the copy engine.

 

[kimchi-tan]

There's a new "kind of" resampler called utaugrowl.
It's not exactly a resampler. It's more a wrapper of other resamplers that adds a growling flag (w0-100).
A description and the ux.getuploader.com download is in this NND video:
http://www.nicovideo.jp/watch/sm22646234

kimchi-tan explains its use in http://utaforum.net/thread-new-growling-resampler
but let me re-type it here.

How to use:
->Unzip utaugrowl.exe into your resampler directory (usually where utau.exe is).
->Rename utaugrowl.exe into @name_ug.exe. @name is the resampler you want to use as base (and it's in the same folder), for example: resampler_ug.exe, freesamp_ug.exe or oldresampler_ug.exe
->Use that *_ug.exe as resampler for the project properties and add the w flag for the notes you want "growling".

As bad news, it can't work with resamplers that has been accelerated with nmasao1's resampler patcher. We have to use original unmodified resamplers.

It can be used with nmasao1's patcher. I think you read my explanation wrong.

---

Does anyone know which frq file each resampler generates? I tried to use Defoko to see which ones are generated but no new frq. files appear in here voicebank folder. So I just screenshoted Miko's instead since I used her the most.

All I know is that:

• .pmk = TIPS
• .vs4ufrq = vs4u

 

[Karma Drak-pa]

^ I would like to correct that second step because it will not actually work otherwise:

Do not rename utaugrowl at all. What you want to do is make a copy of an engine you already have (be it regular resampler or fresamp or even tn_fnds if you want, whatever engine you want), put it in the main directory with utaugrowl and add _ug to the copy's name. Then when you want to make the growling noise, set utaugrowl.exe as Tool 2, not the copy engine.

Well, modifying utaugrowl's name to [resampler]_ug.exe and using it as tool2 works fine, and following what you said doesn't work, at least on my computer. And on Vocaloid.fr, several people noticed that modifying the resampler's name doesn't work~

@kimchi-tan : resamplers working with WORLD's engine (such as the EFB suite or w4u) generate .DIO, .STAR and .PLATINUM files. (and uspec files are generated by UTAU itself, it's the samples' spectrum).

 

[kimchi-tan]

^ I would like to correct that second step because it will not actually work otherwise:

Do not rename utaugrowl at all. What you want to do is make a copy of an engine you already have (be it regular resampler or fresamp or even tn_fnds if you want, whatever engine you want), put it in the main directory with utaugrowl and add _ug to the copy's name. Then when you want to make the growling noise, set utaugrowl.exe as Tool 2, not the copy engine.

Well, modifying utaugrowl's name to [resampler]_ug.exe and using it as tool2 works fine, and following what you said doesn't work, at least on my computer. And on Vocaloid.fr, several people noticed that modifying the resampler's name doesn't work~

@kimchi-tan : resamplers working with WORLD's engine (such as the EFB suite or w4u) generate .DIO, .STAR and .PLATINUM files. (and uspec files are generated by UTAU itself, it's the samples' spectrum).

All three of them use those or only of of each? EFB-PB is very slow for me and using it to regenerate .frq files by changing the path just makes ordinary .frq files (with Defoko, anyway.)

Btw, you didn't quote me. It was a dummy account.

 

[kimchi-tan]

VS4U is now updated at version 1.11!

Download here: http://ackiesound.ifdef.jp/download.html#vs4u