How to Edit .svp File to Convert Head (Falsetto) Voice to Chest (Belting) Voice?

I’m working on this demo song (project file attached), and using An Xiao, Jun, and other male voices.

The problem is, starting from E4 note and up, the singing becomes head voice, also known as falsetto, which is very weak, thin, and sounds bad.

However, when I tried to edit the vocal mode by increasing the “Power” slider to 50, 100, 150, then An Xiao starts to sing in the chest voice, also known as belting, which is strong, solid, and sounds great.

The question is, how can I increase the “Power” slider in the vocal mode passing the maximum 150? Can I edit the project file to achieve 200 or 250?

With “Power” set to 150, An Xiao can sing in his normal chest voice up to B4. When it reaches C5, he starts to sing in a windy head voice.

I wish I could increase the “Power” value to push him to sing up to a solid F5.

Here is the project file.

Project.zip (523.0 KB)

from a thread on this topic about increasing the power of vocals. it seems some voices can go more, and many cannot.

oops, ok, just read the entire thread and you’re on it already.

Inside your project.zip it is not the screenshot with english lyrics but 2 files An Xiao.svp & Jun.svp with no “Yeah” as you mentioned.
But I tried to put a 250 percent inside a svp file and reload it. But no effect ! It seems to be limited to 150 (as you said) by the engine.

Yes I searched and read the entire forum yesterday before I made my post.

In that post you referenced, people were talking about “power” as to “powerful voice”. However, what I meant for “power”, is a vocal mode parameter slider, of which tied to the threshold value of falsetto/head voice being triggered.

In the words, the higher this value is set, the upper note when falsetto is being triggered.

The biggest question is, how to set this value to pass the maximum 150 restricted by the GUI?

It seems doable. Many people from this forum have done it before.

Here is the quote from the original poster.

Listening to other covers makes me realize it’s possible but I’m not sure how anymore. I once read about being able to edit your SVP as a notepad file to manually enter vocal mode parameters and go over the limit of 150 but I’m not sure if that’s something people commonly do or if it was some kind of bug that’s patched.

It’s just a matter of how to find that post on teaching manually edit the parameter in svp without GUI reverting back.

It used to be possible, but it resulted in very obvious synthesis artifacts and was considered a bug so Dreamtonics changed it to revert to 150 if the save file was tampered with.

It is no longer possible in any recent version of the software.

1 Like

Sou desu ne Claire!

Do you remember what the last version was when you could still tamper with the svp without reverting?

I’m using the good old Studio Pro 1.9.0 on a Linux Mint machine, so I may still have a chance to do so.

Also, “artifact” is quite a subjective thing, isn’t it? A user can simply find the highest possible number before major artifact kicks in, and stay on that number. A max of 150 Power is way too conservative, and is wasting so much belting room above.

In my project attached, a Power value of 150 is still solid and clear. I’m pretty sure An Xiao’s training dataset includes sufficient chest/belting voice on high notes, but is being wasted by the falsetto threshold throttle.

Sadly, if this issue can’t be resolved via raising the threshold and removing the throttle, all my male voicebanks will go to the thrash bin.

What a waste of money!

Seems a shame to loose THAT functionality?

A better approach surely to LEAVE the .svp file editable with the Caveat that artifacts MAY result?

THEN the decision to proceed lies with the user - and NOBODY is upset?

Just a thought

The 150 maximum definitely feels arbitrary, and many people don’t care about certain kinds of artifacts if they’re already going for a heavily synthesized/processed sound.

They also made it easier to access extreme ranges of other parameters like tension, breathiness, and tone shift, so it’s odd that the vocal mode change seems at odds with that.

I think it was version 1.8.0 where the fix was implemented, so you’d have to downgrade all the way to 1.7.1 and lose a whole bunch of other features in doing so (like MIDI controller support, ARA, per-note pitch modes, compatibility with any voice database released since November 2022, etc.).

Claire I’m currently in the heat water!

I am fighting myself as to whether or not downgrade to 1.7.1 to make male great again.

To be honest, the only three features I ever need, are Auto Pitch Generation, Instant Mode, and AI Retake. Nothing else is useful to me.

1.7.1 seems to be a perfect candidate to consider, in my case.

Also, 1.7.1 is using HDVM, and 1.8.0 is changing to DPM.

Is there a huge difference in vocal quality, pitch realism, and naturalness, between the two AI algorithms?

Synthesizer V 1.8.0 = Stable Diffusion = Real Human = Number One?

Synthesizer V Studio 1.8.0b1 achieves an exceptional level of realism. In the research phase, Dreamtonics conducted extensive subjective listening tests to compare the new technology against the previous generations and human performance. In a test involving 20 participants designed to evaluate the timbre generation capabilities, statistically significant improvements from the previous generation models were found. In the same test, no statistically significant difference was found between DPM-rendered singing and the performance by humans.