[rakmanenuff]

A question:

Vocal transformer plugins separate the pitched parts from the
formants and treat them separately.

Is there a way to generate audio files that are formant only,
with all the pitch taken out, or vice versa?

[Ethan Winer]

Is there a way to generate audio files that are formant only, with all the
pitch taken out, or vice versa?

No, because a formant is a resonant boost, typically at several frequencies
at once. So you can't "keep" a formant away from the sound any more than you
could keep an EQ boost on a track that is silent.

--Ethan

[Chris Whealy]

rakmanenuff wrote:
A question:

Vocal transformer plugins separate the pitched parts from the formants and treat them separately.


Yes - with varying degrees of success.
Melodyne is one of the best at doing this

Is there a way to generate audio files that are formant only, with all the pitch taken out, or vice versa?

In phonetics, formants are specific overtones in speech or singing that
allow us to distinguish one vowel sound from another, and as such, are
just as "pitched" as the fundamental frequency being spoken or sung.
In order to distinguish vowel sounds correctly, the fundamental must be
lower in pitch than the first formant frequency, otherwise the vowel
information becomes ambiguous. This is why its hard to understand the
words sung by sopranos - they're singing a fundamental that is higher in
pitch that the formant required to determine the vowel information.

In acoustics, formants are the resonant frequencies added to a signal by
(typically) a musical instrument. If you're talking about the overtones
in the sound produced by an acoustic instrument, then it would be less
ambiguous to talk of "timbre" rather than "formants".

It sounds like you're actually asking for two different things he
1) A tool that can subtract the fundamental from a signal and leave you
just with the overtones.
2) A tool that can add overtones to a fundamental frequency

The answer to question 1) is that it should be relatively straight
forward task to perform, but I couldn't say what value such a tool would
bring
The answer to question 2) is yes - get yourself an old analogue
synthesizer and start playing around!

Chris W

--
The voice of ignorance speaks loud and long,
But the words of the wise are quiet and few.
---

[rakmanenuff]

On Oct 21, 11:15*am, Chris Whealy wrote:
rakmanenuff wrote:
A question:

Vocal transformer plugins separate the pitched parts from the formants and treat them separately.

Yes - with varying degrees of success.
Melodyne is one of the best at doing this

Is there a way to generate audio files that are formant only, with all the pitch taken out, or vice versa?

In phonetics, formants are specific overtones in speech or singing that
allow us to distinguish one vowel sound from another, and as such, are
just as "pitched" as the fundamental frequency being spoken or sung.
In order to distinguish vowel sounds correctly, the fundamental must be
lower in pitch than the first formant frequency, otherwise the vowel
information becomes ambiguous. *This is why its hard to understand the
words sung by sopranos - they're singing a fundamental that is higher in
pitch that the formant required to determine the vowel information.

In acoustics, formants are the resonant frequencies added to a signal by
(typically) a musical instrument. *If you're talking about the overtones
in the sound produced by an acoustic instrument, then it would be less
ambiguous to talk of "timbre" rather than "formants".

It sounds like you're actually asking for two different things he
*1) A tool that can subtract the fundamental from a signal and leave you
just with the overtones.
*2) A tool that can add overtones to a fundamental frequency

The answer to question 1) is that it should be relatively straight
forward task to perform, but I couldn't say what value such a tool would
bring
The answer to question 2) is yes - get yourself an old analogue
synthesizer and start playing around!

Chris W

Thanks, both of you.

Sorry if my question was a bit vague. The information
on plugins and what they actually do is generally quite
vague and lacking in detail, and my understanding
of physics/maths/acoustics etc is limited.
Your replies are much appreciated.

What I'd like to do, ideally, is to keep transients/consonants
from a low pitch recording and long notes/vowels
from a re-recording in a higher key. Of course I can
ask the singer to retain the weight and punch of
the lower key whilst singing in a higher key, but I'm
not sure if that's natural or possible.

I'm not happy with what the pitch engine in Logic 8
is giving me on this particular occasion.
I like what the Vocal Transformer plugin is doing
but it sounds too artificial.
I'm familiar with Melodyne but I doubt that it could give
me what I want on THIS particular occasion.

I could also attempt to do some really precise editing
or accept that you win some, you lose some.
Or, use the vocal take in the higher key, then lower the
formants slightly to give it more weight.

Another potential application would be to keep the
transients from someone with really strong, clear, heavy
diction, like a rapper or session singer, and keep
the tonal characteristics/wovel sounds of a singer with weak
attack/diction but nice tone. Or, indeed speak/rap the
words clearly like you would into a vocoder, then
do a separate take focusing more on the notes
and less on the clear diction, and finally layer the two.

[Laurence Payne[_2_]]

On Tue, 21 Oct 2008 11:15:10 +0100, Chris Whealy
wrote:

This is why its hard to understand the
words sung by sopranos - they're singing a fundamental that is higher in
pitch that the formant required to determine the vowel information.

Oh, is THAT why? :-) I thought they were just sloppy over diction and
swapped vowel sounds to flatter the voice, never mind the meaning!

I've just recorded an operatic soprano who thought she'd try her hand
at some lighter stuff. "In the still of the night" became "In the
steel...". I could just about stand that, but when she got to
"...moon growing deem, on the reem of the heel..." I started screaming
:-)

[Chris Whealy]

Laurence Payne wrote:
On Tue, 21 Oct 2008 11:15:10 +0100, Chris Whealy
wrote:


This is why its hard to understand the
words sung by sopranos - they're singing a fundamental that is higher in
pitch that the formant required to determine the vowel information.


Oh, is THAT why? :-) I thought they were just sloppy over diction and
swapped vowel sounds to flatter the voice, never mind the meaning!

I've just recorded an operatic soprano who thought she'd try her hand
at some lighter stuff. "In the still of the night" became "In the
steel...". I could just about stand that, but when she got to
"...moon growing deem, on the reem of the heel..." I started screaming
:-)

Sounds like a Bel Canto trained singer. I also find this /really/
annoying because that school of singing places all the emphasis on the
sound, and almost no emphasis on the words or diction.

If I hear a song, I want to understand the words that are being sung -
not just hear that words are being sung!

This is why I like Emma Kirkby's style of singing - she sings
beautifully, but the words and communication are her top priority.

Chris W

--
The voice of ignorance speaks loud and long,
But the words of the wise are quiet and few.
---

[Don Pearce]

Chris Whealy wrote:
Laurence Payne wrote:
On Tue, 21 Oct 2008 11:15:10 +0100, Chris Whealy
wrote:


This is why its hard to understand the words sung by sopranos -
they're singing a fundamental that is higher in pitch that the
formant required to determine the vowel information.


Oh, is THAT why? :-) I thought they were just sloppy over diction and
swapped vowel sounds to flatter the voice, never mind the meaning!

I've just recorded an operatic soprano who thought she'd try her hand
at some lighter stuff. "In the still of the night" became "In the
steel...". I could just about stand that, but when she got to
"...moon growing deem, on the reem of the heel..." I started screaming
:-)

Sounds like a Bel Canto trained singer. I also find this /really/
annoying because that school of singing places all the emphasis on the
sound, and almost no emphasis on the words or diction.

If I hear a song, I want to understand the words that are being sung -
not just hear that words are being sung!


Most obvious case is the Freddie Mercury / Monsarrat Caballe version of
Barcelona. Mercury is powerful and meticulous in his diction while
Caballe just makes noises which aren't even timed particularly well.

This is why I like Emma Kirkby's style of singing - she sings
beautifully, but the words and communication are her top priority.
Chris W


Hear hear to this!

d

[Chris Whealy]

rakmanenuff wrote:
What I'd like to do, ideally, is to keep transients/consonants from a low pitch recording and long notes/vowels from a re-recording in a higher key.
There are two fundamental problems he
1) Consonants also contain formants
2) A consonant followed by a vowel is always blended into single unit
of speech.

Just for a laugh, try saying a voiced consonant such "d" or "b", but
suppress your automatic urge to add a subsequent vowel sound.
If you can achieve this, what sound do you end up with? Is it
discernible as a distinct consonant or just a noise that seems to form
no part of speech?

Of course I can ask the singer to retain the weight and punch of the lower key whilst singing in a higher key, but I'm not sure if that's natural or possible.

Wouldn't it then follow that if you tried this exercise artificially,
that it would also be either impossible or unnatural? :-)
I'm not happy with what the pitch engine in Logic 8 is giving me on this particular occasion.

Not surprising
I like what the Vocal Transformer plugin is doing but it sounds too artificial.

Also, not surprising
I'm familiar with Melodyne but I doubt that it could give me what I want on THIS particular occasion.

Given the options of available tools and the task at hand, I think
Melodyne is your best option. Do you have the Studio version?
I could also attempt to do some really precise editing
??? Using what tool?
or accept that you win some, you lose some.

Yup
Or, use the vocal take in the higher key, then lower the formants slightly to give it more weight.

You'll need to be very careful here. It is easy to produce a very
unnatural sound if you start noodling around with just the formants.
Another potential application would be to keep the transients from someone with really strong, clear, heavy diction, like a rapper or session singer, and keep the tonal characteristics/wovel sounds of a singer with weak attack/diction but nice tone. Or, indeed speak/rap the words clearly like you would into a vocoder, then do a separate take focusing more on the notes and less on the clear diction, and finally layer the two.


Good luck!!

The problem here is that in addition to the formants carrying the vowel
information, their formation is specific to the shape of the individual
singers' throat, larynx and chest. These physical differences are what
give us our characteristic voice sounds.

If you then combined elements of speech from two different people, you
could end up with a sound that will be recognisable as speech, but might
loose the distinct attributes that allow the listener to perceive it as
male or female.

To be honest, this just sounds like alot of hard work to me. In terms
of time/effort/cost, I think it would be quicker (and much simpler) to
find a vocalist who can sing the part the way you want.

Chris W

--
The voice of ignorance speaks loud and long,
But the words of the wise are quiet and few.
---

[Chris Whealy]

Don Pearce wrote:

Most obvious case is the Freddie Mercury / Monsarrat Caballe version
of Barcelona. Mercury is powerful and meticulous in his diction while
Caballe just makes noises which aren't even timed particularly well.
Ergo...

Monsarrat Caballe - Bel Canto
Freddie Mercury - Can Belto

:-)

Chris W

--
The voice of ignorance speaks loud and long,
But the words of the wise are quiet and few.
---

[Laurence Payne[_2_]]

On Tue, 21 Oct 2008 15:53:07 +0100, Don Pearce
wrote:

Most obvious case is the Freddie Mercury / Monsarrat Caballe version of
Barcelona. Mercury is powerful and meticulous in his diction while
Caballe just makes noises which aren't even timed particularly well.

Or that recording of "West Side Story" with the opera voices. Apart
from no-one (except the Krupke boys) coping with the rhythms, they
cast a Hispanic-sounding Tony. I HOPE Bernstein just saw it as a big,
lucrative joke.