Kyma Forum
  Tips & Techniques
  Vocalize

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Vocalize
BardetSamy
Member
posted 30 September 2010 17:25         Edit/Delete Message   Reply w/Quote
Hi,
My next work with the Kyma is a bit difficult... i must make a monkey talk
The last days i began trying the memory writer to make something like JED do on his site with a camera's sound ( http://jedsound.com/blog/?p=18 ),but without success.
I don't manage to vocalize a sound of a monkey like that with the memory writer, but i think it's the more realistic way to do it...
If someone have an idea ?
Thanks
Samy


IP: Logged

SSC
Administrator
posted 02 October 2010 16:39         Edit/Delete Message   Reply w/Quote
What kind of monkey?

IP: Logged

BardetSamy
Member
posted 05 October 2010 02:33         Edit/Delete Message   Reply w/Quote
Hi,

Thanks for reply SSC
It's a mix between a chimpanzee and a baboon.
I search to make something very realistic, if a monkey talking can be realistic
I want to vocalize this animal with my voice so i began with vocoder but seems to be too synthetics. Now i'm learning the prototype tracking...
And i received the mail with the link of Jed's site and his approach of vocalize intrigued me. I tried to do the same with my monkey but i must admit that i would need a big training lol

Best regards,

Samy

IP: Logged

robertjarvis
Member
posted 05 October 2010 05:01         Edit/Delete Message   Reply w/Quote
Have you checked out the ModalFilter Prototype?

IP: Logged

SSC
Administrator
posted 05 October 2010 08:36         Edit/Delete Message   Reply w/Quote
Does it have to have recognizable words? Or just expressive pitch?

For recognizable words, you might try the RE analysis. But you would need some screechy, broad band sounds from the monkey as input to excite the filter.

IP: Logged

BardetSamy
Member
posted 05 October 2010 11:41         Edit/Delete Message   Reply w/Quote
Hi,

Thanks robertjarvis for the idea but what is this new prototype again ?
I will test later...
SSC-> yes it must be some recognizable words, so i will work the RE analysis, thanks.

SAmy

[This message has been edited by BardetSamy (edited 05 October 2010).]

IP: Logged

SSC
Administrator
posted 05 October 2010 16:17         Edit/Delete Message   Reply w/Quote
I would be happy to try some experiments if you have some samples of the monkey cries that you could share.

IP: Logged

BardetSamy
Member
posted 06 October 2010 12:37         Edit/Delete Message   Reply w/Quote

Chimpanzee.wav


Chimpanzee2.wav

 
With pleasure

I will send the monkey samples tommorrow.

Thanks


maj : you can find 2 samples of chimpanzee... very impatient to hear your creations.

[This message has been edited by BardetSamy (edited 07 October 2010).]

IP: Logged

BardetSamy
Member
posted 25 October 2010 14:34         Edit/Delete Message   Reply w/Quote
My investigation continue...
I saw the image with the monkey and the good news is that he don't speak really, he have some expressions laugh, angry, etc...
So i will try different approach for that.

IP: Logged

SSC
Administrator
posted 26 October 2010 23:57         Edit/Delete Message   Reply w/Quote

chimpspeak.zip

 
Here's a chimp you can play using the Wacom Tablet (eraser end) or Kyma Control keyboard (with no rounding) or on the Continuum Fingerboard. Control the pitch with left/right and the timbre with forward/backward and you can create some expressive sounding utterances.

IP: Logged

BardetSamy
Member
posted 30 October 2010 07:18         Edit/Delete Message   Reply w/Quote
Thanks !!!
I begin to understand this way to use the spectra analysis... it's fantastic. I tried with other spectra, i will post some results
I will try some expressive sounding utterances.

Samy



IP: Logged

dieffe
Member
posted 03 November 2010 06:07         Edit/Delete Message   Reply w/Quote
Hi
I'm working on something similar: I'm trying to vocalize the trump of an elephant.
I tried different approaches, and by far, with SOS I get the best sound quality. With the Quasi-Harmonic analysis and the oscillator bank, I get lots of artifacts (I'm quite sure that I'm doing something wrong with the analysis).
I tried the TAU (and I found that the TAU can give me all the flexibility I need, morphing and twisting the sound) but the quality is really not on pair with the SOS, it sounds much thinner.

I'm doing something wrong with the PSI analysis or SOS is the best way to go? Any tips to improve the expressivity of the elephant's trump? how could I make it morph between more than two sound (with SOS)?

thanks in advance (and sorry for the newbie questions)

davide

IP: Logged

SSC
Administrator
posted 03 November 2010 09:36         Edit/Delete Message   Reply w/Quote
Do you have an example of the elephant you could upload here? If it is inharmonic and includes noise elements, then the quasi harmonic analysis and TAU will have artifacts (they work best on harmonic signals).

IP: Logged

photonal
Member
posted 03 November 2010 10:11         Edit/Delete Message   Reply w/Quote
what about getting the elephant to produce a bit more 'umph' or presence in its call?

IP: Logged

dieffe
Member
posted 03 November 2010 11:13         Edit/Delete Message   Reply w/Quote

elef_barr_08.wav


elef_barr_07.wav

 
these are two of those elephant trumpets that I'm using.
I'm trying to make the elephant angry, than i become a pain in the neck for the actors, than steal a bag, etc (it's a slapstick comedy). It already works with the sound editing alone, but I'm trying to improve the "acting" of the elephant, making it more funny.
I like the idea to give it a little "humpf".. have to try something...

with the SOS, now I can gracefully pitch up and down an have a nice time stretch( and make the elephant performance a bit more alive), but I'm looking forward to other things to try

thank you to all

davide

IP: Logged

BardetSamy
Member
posted 10 December 2010 10:11         Edit/Delete Message   Reply w/Quote

TRacking.zip

 
Wow, the topic began by monkey and now we talk about elephant...
That's a great zoo topic
Sorry for this long silence, but each time i want to explore more Kyma a new job arrives...
Thanks again SCC, your sound works great !!!
Can it be possible to manipulate the !Keytimbre parameter of the eraser pen, i mean , perhaps smaller amplitude or drived by a voice...
So i tried with this new experience which is about the frequency, amplitude tracking of a voice to make the monkey speak.
You can find an example attached.
It's not the best sound ever... the sample of the monkey follow very well the amplitude of the voice.
I can modulate the frequency with my voice but it's not very simple...
It would be kind that instead of following only the frequency it could following the vowels formants... perhaps a stupid question ?
I don't know how it can be more close to the voice...

Thanks dieffe for the idea of SOS, i will try this now.

Best

SAmy

[This message has been edited by BardetSamy (edited 11 December 2010).]

IP: Logged

BardetSamy
Member
posted 13 April 2011 13:55         Edit/Delete Message   Reply w/Quote

SingeKyma.zip


Singe.zip

 
Hi Kymaland,

After a long period without Kyma ( it was horrible ), i'm back to explore again the world of sounds !!!!
And to begin i'm working on an animation movie with this famous talking monkey, and it's not simple...
I attach some examples of wav files and a Kyma sound of my work at the moment. But i'm not satisfied enough
I can make some expressions of the monkey but it's not " talking enough ".
I've tried the Tau file to morph the monkey with "reel" words without success...
Then, i tried Re Analysis, but i don't really understand the way of use Re Analysis... for this.
And finally, i wanted "vocolize" the sound like i've talk about at the beginning of this topic with the memory writer, in this way

"This patch allows to adjust the IN and OUT of a sample loop while I’m changing the rate. Kyma records the loop segment at each BPM (adjustable) in the RAM and as the IN and OUT move constantly, I can dial on any part of the sample and make the sound vocalizing."

but without success due to my ignorance...

So if someone can explain some details to me, it would be a pleasure

Samy

[This message has been edited by BardetSamy (edited 13 April 2011).]

IP: Logged

CharlieNorton
Member
posted 13 April 2011 21:05         Edit/Delete Message   Reply w/Quote
Hello Samy

I have a few vague ideas but before I explore them I need a bit more context?

Is there a script for the monkey? Does the monkey have actual words or is it supposed to communicate in utterances/implied meaning?

How long is it?

Will you have a video to work to?

I can see two main approaches. A) Take a human voice and manipulate it with monkey formantz/frequencies or B) Take recordings of monkeys arrange them to the video so they inflect and sync, then vocode or morph the speech to intelligence.

The big challenge is the amount of dialogue that needs creating. If it is long, then finding some real time way to act/perform the speech would be helpful.

If there is only a limited amount of dialogue then you could possibly sculpt each segment in TAU or create analysis of each segment.

I think the realities are thus: To pull this off you need A) The finished video B) whole load of human attempts at the vocalisation/speech C) Bucket full of monkey recordings D) an epic amount of time.

This does sound like an interesting but formidable challenge.

Charlie

IP: Logged

BardetSamy
Member
posted 14 April 2011 01:39         Edit/Delete Message   Reply w/Quote
Thanks Charlie for the reply

I will explain a little more for help :
the monkey communicate in utterances/implied meaning in general but have
a few sentence in the movie.
I've already synchronize some sounds on the video, it's not so long.
I does agree with your B) solution :
"Take recordings of monkeys arrange them to the video so they inflect and sync, then vocode or morph the speech to intelligence."

but i've tried to morph them in the Tau editor but without success and the
time i've used vocoder it's sound "synthetic" .

It's a very interesting challenge and i would like to show the
power of the Kyma for this !!!

Best,

Samy

[This message has been edited by BardetSamy (edited 14 April 2011).]

IP: Logged

CharlieNorton
Member
posted 14 April 2011 10:16         Edit/Delete Message   Reply w/Quote
Samy,

Ok so if we consider Tau morphing for the moment as a focus, my thoughts lead to this:

In my experience of morphing, there are two ways of achieving results, A) morphing completely between two sources. In my opinion, this does not have the subtlety even at a half way setting.

B) Experiment with the sounds and find one that allows you to morph the frequency, time, formant etc independently, it is thought careful experimentation and combination of parameter selection that you might find what you are looking for? Formant of the human, yet the frequency of the monkey or the other way around?

See what I am getting at? You may have considered this before.....

Good Luck

Charlie


IP: Logged

BardetSamy
Member
posted 15 April 2011 14:13         Edit/Delete Message   Reply w/Quote
Charlie,

Thanks for your "encouragement"

It's true that the complete morph between 2 samples is not simple...
But like you've said the source sound of the morph is very important.
After an afternoon using the Tau editor i approach a great result,
only sculpting the sounds but not morphing them...

Best,

SAmy


IP: Logged

SSC
Administrator
posted 15 April 2011 16:52         Edit/Delete Message   Reply w/Quote
Hi Samy,

I wonder if your attached example might sound more animal-like with a continuous keyboard.? I tried playing your resynthesis from Kyma Control, but I think it could be even more expressive using the Wacom tablet or a Continuum (since they both have continuous amplitude control as well as pitch).

This example is not great, but I think it might get a little closer to speech-like sounds when you can have continuous pitch glides in addition to some discrete steps?

IP: Logged

BardetSamy
Member
posted 16 April 2011 01:10         Edit/Delete Message   Reply w/Quote
Hi SSC,

Wowww !!!! How do you manage to this ??? It's my sound with the Wacom tablet ???
I does agree about my sound, when i listen it today it seems weird
I tried several times with the Wacom tablet to do like your sound at 25s, more speaking, without success.... perharps possible problem in my settings of the wacom. I will explore it with my new sounds from the Tau editor.
But i can say thanks to you, that it's possible to do it with Kyma !!!

Greats Thanks SSC,

SAMy

[This message has been edited by BardetSamy (edited 16 April 2011).]

IP: Logged

BardetSamy
Member
posted 03 May 2011 11:25         Edit/Delete Message   Reply w/Quote

SingeMorphing.zip

 
Hi,


I still have not managed to make a monkey talk but I managed to make it sing, see the attached file...
My first great morphing, i'm proud of it

Excuse me SSC but i don't understand this words :

"I think it might get a little closer to speech-like sounds when you can have continuous pitch glides in addition to some discrete steps? "

i really need to make it talk like you did , but each time i try it with the wacom tablet it loose the attack...

Thanks again for your help,

Best

SAmy


IP: Logged

SSC
Administrator
posted 04 May 2011 07:18         Edit/Delete Message   Reply w/Quote
Amazing! (also hilarious!

(In answer to your question, I was just suggesting making short pitch glides with some stops and gaps to imitate speech. Regarding losing the attack, do you have an envelope on the resynthesis? It might be that the attack on the original source is naturally slow so that you do not hear it reach the voiced part when you do short gestures on the Wacom? )

IP: Logged

BardetSamy
Member
posted 05 May 2011 02:14         Edit/Delete Message   Reply w/Quote
Thanks SSC,

Yes, he's hilarious this monkey...

I'v tried again with the wacom tablet and with my resynthesis of monkey and a nooby question comes suddenly : what can give me more to play a resynthesis instead a sample ? ( in that case of a small sample )

I love the Tau Player.

Best,

Samy

[This message has been edited by BardetSamy (edited 01 July 2011).]

IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c