Back then, I always wondered how UTAU got the logical ranges for each UTAUloid. I later found out about generating frequency files and someone saying that fresamp generates more stable frequencies compared to the default resampler so I regenerated some of my UTAUloids's FRQ files.
Here's how to regenerate your UTAUloid's FRQ files (sorry about the cropping, but it's still understandable):
I noticed a difference in the logical ranges. Here are shots of side-by-side comparisons (unregenerated on the left and fresamp regenerated on the right):
[size=-5]Okay, that last one's a joke (yeah, I have Tree XD) but that's seriously what fresamp gave me XD[/size] This lowered the uses percentage. I think this is because of lacking pronunciations like
"nga" "nge" "ngi" "ngo" and "ngu" (Defoko has these and after fresamp regenerating her, it's still at 100%)
So it's my conclusion that the FRQ files affect the logical range.
An UTAUloid's logical range can be affected by different ranging appends and and breath files which tend to be very low to some which may make the logical range inaccurate, so regenerating FRQ files with fresamp for getting a more accurate logical range may not always work (although it improves the stability of the FRQ files).
Some UTAUloids' samples may be already generated with fresamp so the logical range may stay the same.
So far, this is the weirdest vocal range of an UTAUloid I have:
I complied all of Miko's appends into her Kire voicebank (you can tell by looking at the change of her label and details). I regenerated her whole bank (it has 5 appends, random samples, ending breath, long tone, staccato vowels, and 3 pitch banks for Kire. It lasted two hours) but it's still the same.
If you're going to regenerate your UTAUloids's FRQ files, consider the time usage you're going to need:
Here's how to regenerate your UTAUloid's FRQ files (sorry about the cropping, but it's still understandable):
I noticed a difference in the logical ranges. Here are shots of side-by-side comparisons (unregenerated on the left and fresamp regenerated on the right):
[size=-5]Okay, that last one's a joke (yeah, I have Tree XD) but that's seriously what fresamp gave me XD[/size] This lowered the uses percentage. I think this is because of lacking pronunciations like
"nga" "nge" "ngi" "ngo" and "ngu" (Defoko has these and after fresamp regenerating her, it's still at 100%)
So it's my conclusion that the FRQ files affect the logical range.
An UTAUloid's logical range can be affected by different ranging appends and and breath files which tend to be very low to some which may make the logical range inaccurate, so regenerating FRQ files with fresamp for getting a more accurate logical range may not always work (although it improves the stability of the FRQ files).
Some UTAUloids' samples may be already generated with fresamp so the logical range may stay the same.
So far, this is the weirdest vocal range of an UTAUloid I have:
I complied all of Miko's appends into her Kire voicebank (you can tell by looking at the change of her label and details). I regenerated her whole bank (it has 5 appends, random samples, ending breath, long tone, staccato vowels, and 3 pitch banks for Kire. It lasted two hours) but it's still the same.
If you're going to regenerate your UTAUloids's FRQ files, consider the time usage you're going to need:
- Regenerating FRQ files on a single-tone CV voicebank (if the resampler patcher is used on fresamp) can only take a minute long.
- Regenerating FRQ files on a single-tone VCV voicebank (if the resampler patcher is used on fresamp) can take about 10-15 minutes.
- The speed of fresamp may also depend if your computer is fast or not
![[K] of Honor](/data/medal/1_1406930084l.jpg)
I never actually thought of this as a tutorial; just something to share. (and then I realized that I kept updating this post with more information than I thought OTL)