Real-Time Formant Shifting: Any plugins do this? (HELP?!)

[deafboy]

I have a vocal track where I want to shift the entire formant of the track way down.

(I'm ideally looking for a real-time formant shifter, so that I can audition the takes immediately after recording them.)

I know of some formant filters, but I just am looking for something really simply that just shifts the formant of the entire sound.

Since I want to do this on a vocal track, I don't want noticable distortion AND I don't want the high-frequencies to be noticably compromised.

SOMEONE HELP!!!!

[shamann]

xoxos made one in synthedit, but its more proof-of-concept, as it pops and clicks a lot. Smoky joe is what it is called, the source was available if you wanted to have a go at improving.

I must admit, I've never fully understood the concept of a formant shifter against a pitch shifter. Any good descriptions out there that you know of? Google has always proved fruitless (and tends only to point to posts around the web that xoxos has made looking for said same processor.)

Also Leapfrog Rephrase has a formant function, but I'm not sure if it is a formant filter or a shift. There's a really expensive one out there, but can't remember who made it.

[Hovmod]

Isn't this what Melodyne does well? Out of my price range, but it sure looked tasty...

[deafboy]

Isn't this what Melodyne does well? Out of my price range, but it sure looked tasty...

I don't believe that program is real-time.

[useruseruser]

What do you understand as a "formant shifter"?

People often mismatch formant filters with formant shifters.

I would call formant shifting always in close connection with pitch shifting. As the mechanism, which "shifts" the formant characteristics with the amount of pitch change.

That means finally, that the well known Mikey Mouse and Darth Vedder effects are prevented as much as possible.

But such an implementation just could do the contradiction: shifting the formants only, but preserve the pitch. That is also known as "changing the gender" of a voice.

Well, there is currently another thread, where I offered a to build a plugin on base of PSOLA. But the interest was quite low.

That's an algorithm which is specialized to exactly do that. It (Pitch Synchronuous Add and Overlapp) is still limited. It works merely good with dry and clean monophonic voises, like a single singer
without any effects. It dosn't work with polyphonic material and needs a well defined frequency range.

It's still able to process realtime, but then with limited quality and reduced feature set. Althought, slowing down and speeding up voices is never the less only possible with offline processing.

[Hovmod]

Isn't this what Melodyne does well? Out of my price range, but it sure looked tasty...

I don't believe that program is real-time.

You might be right. It's certainly not real-cheap, so I won't find out.
Good luck, and keep us posted...

[deafboy]

What do you understand as a "formant shifter"?

I don't want to alter the pitch at all.

I only want to alter the formant.

I didn't ask for formant filter plugins because most of those plugins seem really complicated and alter really specific aspects of the formant, rather than the entire thing.

There is another thread on formant filters, but those plugins seemed to mess with the formant in weird ways, while I ONLY want to shift the entire formant down.

[shamann]

Arboretum Hyperprism was the expensive one I was thinking about. DX only though, not sure if it's real-time or not.

J&H, I'd certainly be interested in seeing your PSOLA-based plugin.

[xoxos]

What do you understand as a "formant shifter"?

I don't want to alter the pitch at all.

I only want to alter the formant.

i think i got tired of nudging for this for years a few years ago.. :p

j&h if you can up the quality on 'smoky joe' for a f/w formant shifter, coulja do it?? :p

deafboy - dont take any shit. if you say 'formant shifter,' people will chat the thread into oblivion with "why, do you mean formant scaling,f ormant fransposition, shifting, et c." when somehow the simple "like my pitch shifter but w/o te pitch please" is NEVER understood ever ever ever ever ever

deaf deaf deaf

[useruseruser]

Well. seeing don't help that much. Aditioning was better.

[useruseruser]

What do you understand as a "formant shifter"?

I don't want to alter the pitch at all.

I only want to alter the formant.

I didn't ask for formant filter plugins because most of those plugins seem really complicated and alter really specific aspects of the formant, rather than the entire thing.

There is another thread on formant filters, but those plugins seemed to mess with the formant in weird ways, while I ONLY want to shift the entire formant down.

So then you want to change the gender?

[useruseruser]

I mean, if you actually want to apply only very subtile formant modifications, so then I suggest to experiment with formant filters or specialized formant tools.

If you are interested to apply formant shifting with PSOLA technology, which shifts the formants proportional on the entire frequency content (also known as "change gender"), then you can wait for my free tool VoicePitch (monophonic material), which is the counterpart of my allready available free Tool AudioPitch (polyphonic material without formant shifting).

It will be basicly a MIDI enabled pitch shifter effect, which has the possibility to shift the formants separately. As already stated, it works well only with monophonic voices in limited frequency range (i.e. dry recorded solistic instruments or singers).

[Lunch Money]

I'd be interested, too. I'd just want to make my own voice sound subtly like somebody else's, so that I can do my own harmonies (not necessarily shifting pitch, I'll sing to pitch myself) without sounding like I'm just singing with my clones.

Greg

[useruseruser]

Can anyone point me to a dry recorded singing voice, which I can use to demonstrate my effect (before I start to compile a plugin for distribution, because that's allot of work, you know)?

[semiquaver]

melodyne will work in almost real time. A VST bridge pipes the audio in - you then click the "analyse button". Then you simply play back your track - as it plays back you can vary the formants using the "global playback offsets". pretty darn groovy.