I need tips for recording a utau

Sync_Ye

Ruko's Ruffians
Defender of Defoko
I'm thinking of creating a Utau, but I need some tips like for example: bpm ideal for VCV japanese and if you edit and put some filter on the voice samples could be something good. (I don't need Reclist for this because I already have a very good one)

Here I have an example of a voice bank that I made just to test the same:

Tone: C#3 - 100 bpm

It the only edits I did was to remove the noise of the recordings, and add an equalizer that removes some frequencies of my choice, not done any oto for he has only to test, so I left the oto of the moresampler 0.8.3 in romaji.
Note: Didn't know where to put this so if in wrong place please forgive me, and if something is written, strangely, was the translator
 
  • Like
Reactions: Melomad and partial

WilyemAndK

✧・゚: *✧・゚:* professionally slow *:・゚✧*:・゚✧
Retired User
Defender of Defoko
That sample sound pretty neat. I'd love to see how that bank turns out!

A couple of tips would be just to record at a comfortable pitch and speed. Kasane Teto's VCV voice bank is recorded at a pretty average tempo, so you could record is around that speed perhaps.
Also, your UTAU has a very thick accent from what I see. That's good in some cases, but when singing in Engrish and other languages by just using the Japnese bank, it might sound a bit strange.
And remember: if you can't pronounce the Japanese R sound (the half rolled/trilled R specifically), then pronounce it as an L sound. If you know how to pronounce the R sound, which does take a bit of work to get used to, then I recommend recording both the original Japanese R sound AND an L sound, aliased under R except with a 2. For example, you'd have stuff like -ra, -re, -ri, -ro, and -ru, but you'd also have -ra2, -re2, -ri2, -ro2, -ru2, and so on so forth. The reason I recommend recording both of those is because when your UTAU is singing faster songs, it'll sound weird with the Japanese R sound all the time.

I hope these tips help you!
-Wilyem
 
  • Like
Reactions: Sync_Ye

Sync_Ye

Ruko's Ruffians
Defender of Defoko
Thread starter
That sample sound pretty neat. I'd love to see how that bank turns out!

A couple of tips would be just to record at a comfortable pitch and speed. Kasane Teto's VCV voice bank is recorded at a pretty average tempo, so you could record is around that speed perhaps.
Also, your UTAU has a very thick accent from what I see. That's good in some cases, but when singing in Engrish and other languages by just using the Japnese bank, it might sound a bit strange.
And remember: if you can't pronounce the Japanese R sound (the half rolled/trilled R specifically), then pronounce it as an L sound. If you know how to pronounce the R sound, which does take a bit of work to get used to, then I recommend recording both the original Japanese R sound AND an L sound, aliased under R except with a 2. For example, you'd have stuff like -ra, -re, -ri, -ro, and -ru, but you'd also have -ra2, -re2, -ri2, -ro2, -ru2, and so on so forth. The reason I recommend recording both of those is because when your UTAU is singing faster songs, it'll sound weird with the Japanese R sound all the time.

I hope these tips help you!
-Wilyem
thanks for the tips, actually I already recorded samples "L" and "R" separate (only that this R used is the same as the Portuguese of brazil because I asked to many native japanese, and they spoke that this R is the same as they use more the moresampler you have configured them in a way very strange from what I saw here)
 
  • Like
Reactions: WilyemAndK

WilyemAndK

✧・゚: *✧・゚:* professionally slow *:・゚✧*:・゚✧
Retired User
Defender of Defoko
thanks for the tips, actually I already recorded samples "L" and "R" separate (only that this R used is the same as the Portuguese of brazil because I asked to many native japanese, and they spoke that this R is the same as they use more the moresampler you have configured them in a way very strange from what I saw here)
Great! I'd love to see your voicebank once it's complete!
 
  • Like
Reactions: Sync_Ye

Melomad

Ritsu's Renegades
Defender of Defoko
I'm thinking of creating a Utau, but I need some tips like for example: bpm ideal for VCV japanese and if you edit and put some filter on the voice samples could be something good. (I don't need Reclist for this because I already have a very good one)

Here I have an example of a voice bank that I made just to test the same:

Tone: C#3 - 100 bpm

It the only edits I did was to remove the noise of the recordings, and add an equalizer that removes some frequencies of my choice, not done any oto for he has only to test, so I left the oto of the moresampler 0.8.3 in romaji.
Note: Didn't know where to put this so if in wrong place please forgive me, and if something is written, strangely, was the translator

As for the bpm, there is no ideals one but older JPN voicebanks tend to be recorded with 100bpm and newer one on 120/130bpm.
If I were you I would not remove frequencied via EQ in the sample because it's hard work for nothing. If people want to cut the low end frequencies of your UTAU they do it in mixing. Doing it directly into the sample could result in a UTAU wich is hard to mix.
 

Sync_Ye

Ruko's Ruffians
Defender of Defoko
Thread starter
As for the bpm, there is no ideals one but older JPN voicebanks tend to be recorded with 100bpm and newer one on 120/130bpm.
If I were you I would not remove frequencied via EQ in the sample because it's hard work for nothing. If people want to cut the low end frequencies of your UTAU they do it in mixing. Doing it directly into the sample could result in a UTAU wich is hard to mix.
thank you I will take this into consideration, the frequencies that I removed were equivalent to the noise (you have no idea how brazil is noisy where I live), I record the samples all in a single file and use a program that cuts and appoints the samples with times equal
 
  • Like
Reactions: Melomad

Similar threads