Jump to content


Photo

[Instruments] OopsVocoder

vocoder native instrument robot voice

  • Please log in to reply
13 replies to this topic

#1 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 22 February 2017 - 16:03

File Name: OopsVocoder

File Submitter: OopsIFly

File Submitted: 22 Feb 2017

File Category: Instruments

License: CC0, Public Domain


Version 1.1

 

Ok, so I uploaded a slightly enhanced version. The vocoder is now a lot more intellegible. I'm myself surprised how it sounds now. The first version sucked. I had forgotten to save the exp instead of linear mapping of the signal followers in between.

 

Also don't take it badly, that the voicings of the vocoder is a bit strange. I extra made some changes to make it sound more gay. You know, I don't care whether a person is actually homosexual or not, its his/her own buisiness. But I do get the kick about gay voice timbres. Especially when tweaking the 2 notable dials of this instrument. And it will generally lisp, in whatver voicing, and I really do get the kicks even more out of this fact. This is on purpose. And also my humble sense of humor. No I won't tune another version for you that won't sound as silly. Though I'd know how to. MWHAAAHAAAHA!

 

But seriously, I first wanted to tune this bugger further and make it more powerful and configurable, but now I got around studieing a bit of relevant info, that will allow me to hopefully make e much better device.

 

 

Version 1.0

 

As Respond to the bellow's vocoder sound - I did this thing yesterday, and some futile extra tuning today. It is interesting how both of us went into totally different directions regarding the sound.

 

Yes, it is not so very intellegible, earlier versions before I added some extras seemed better. But this also comes because the modulator is only present in the result via filtering, no sound is routed through from it. I'm not so happy about it, maybe I find a muse for tuning it better one day. With scientific sheets on speech intellegibility or so.

 

Also now my ears bleed from hearing about a train to...now where does it go?...all the time over and over until I understood totally different pretty insane things, and not the original words anymore however hard I tried...If you have this while tuning sounds, a pause from the sound of 1-2 hours can bring back the original meaning of the speech.

 

I went for trying to achieve a pseudo analog driven character, trying to keep in in raw but still smooth sound regions. It has some macros to change the sound, play around.

 

Sorry, it takes some cpu because of all the filters etc, I disabled oversampling to make it lighter on the machine.

 

For using this in actual tunes - I personally would delete the speech sample that is in there, and add new ones. Make some keys free of the saw and noise in the upper or lower end of the keyzones, lay your speech samples onto these notes. And route these samples to the mod set and fx chain that the original speech sample goes. Then in pattern or phrase you can compose a melody within the saw/noise keyzones, and trigger the modulator samples via other keys and sample offsets and such to sequence the "speech".

 

Happy vocoding! Yes a VST will sound much better, this is more toying around for the fun of it.

 

Credits for speech sound that comes with this: http://www.freesound...y/sounds/64965/ - cc attribution license, consider my instrument public domain if you remove the sample


Click here to download this file


  • TheBellows and ffx like this

#2 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 26 February 2017 - 00:41

Version 1.1, after some (t)weak extra session, now sounds a bit clearer, but even sillier.



#3 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 02 March 2017 - 01:08

Added silly demo.



#4 danoise

danoise

    Probably More God or Borg Than Human Member

  • Renoise Team
  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • 6250 posts
  • Gender:Male
  • Interests:wildlife + urban trekking

Posted 04 March 2017 - 13:06

Very awesome! IMO it sounds better than a lot of vocoder plugins, and the macros are spot on. 


  • TheBellows likes this

Tracking with Stuff. API wishlist | Soundcloud


#5 gentleclockdivider

gentleclockdivider

    Guruh Motha Fakka is Levitating and Knows Everything About Renoise Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • 2476 posts
  • Gender:Male
  • Location:something like a sea or a river...just moving
  • Interests:slacking

Posted 04 March 2017 - 14:23

Sounds gritty , in a good way .

 

I remember building a vocoder was the  first thing I did when the envelope follower was added ( 2.5 ).

If you offset the carrier bands ,you can achieve some nice fromant shifts .

I made a crystal clear 512 bands resiner/vocoder  ( the modulator is actually a 512 partial  sine osc ) sounds ace 


THERE IS NO RELIGION SUPERIOR TO THE TRUTH


THERE IS NO TRUTH SUPERIOR TO THE ONE I CHOOSE

C'EST MIEUX D'ETRE BELLE ET REBELLE


QUE MOCHE ET REMOCHE



RUBIO RABIA

 


#6 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 06 March 2017 - 23:23

sines for each bin instead of filtering a synth carrier - interesting, never had my eyes on that type of resynthesis...though I've heard it sometimes. Real difficulty is probably tuning the oscs so it sounds sick enough and phasing is just right...

 

But nothing I'd want to try doing in renoise native.

 

In renoise native I like to do simple vocoders.

 

Oh, and don't forget the analog lowpass filter on the modulator voice, before it is fed into the vocoder. With some drive and enough resonance. Frequency 1.8-2.8 Khz stationary. This is how we kick it, ...



#7 gentleclockdivider

gentleclockdivider

    Guruh Motha Fakka is Levitating and Knows Everything About Renoise Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • 2476 posts
  • Gender:Male
  • Location:something like a sea or a river...just moving
  • Interests:slacking

Posted 07 March 2017 - 11:49

same principle really.

Here the resiner =128 partials ( voices ) 

30tjjoy.jpg

https://app.box.com/...l1g3xjk03s0zqn6

The so called formant shift is done by offsetting the carrier frequencies ( partials ) 

 

There's no need to tune the osc's , the modulator bandpass filter are logarithmic values between 1-130 ..which span the whole frequency range.

Each voice in reaktor has a unique i.d. nr.  meaning voice .1 = nr.1 ,voice 2=2 etc ..these numbers are used as mulipliers for each band frequency , the voice module is verry efficient for that .

 Ofsetting the whole lot for the carrier etc ..

Out of experience , I noticed that log.values ( especcially for vocoding ) yield far better results then linear freq.values 


Edited by gentleclockdivider, 07 March 2017 - 12:01.

THERE IS NO RELIGION SUPERIOR TO THE TRUTH


THERE IS NO TRUTH SUPERIOR TO THE ONE I CHOOSE

C'EST MIEUX D'ETRE BELLE ET REBELLE


QUE MOCHE ET REMOCHE



RUBIO RABIA

 


#8 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 07 March 2017 - 16:16

ah, now I geddit. Its using an additive synthesis approach to the carrier, modulating each harmonic. Thing for sure sounds very nice and clean! I first thought about something in the lines of a downsampling phase vocoder resynthesis, or other freaky stuff like this that sounds like a 5 kbps mp3 or so...

 

And yeah, if you look at the oopscoder it does shift the carrier bands (the castration LVL macro...). Just in a rather brainfucked way, making it sound odd, by purpose.

 

I'm currently working on a 10 band native vocoder with better sound, and it already has the clean shifting of carrier bands, thx to the formula device.

 

Also I'm tuning the buggers for speech, not other modulators. For general purpose you just need more bands, the more the better. The 10band version is based on scientific speech intellegibility data, so it will still suck for drums or whatever else you could input into a vocoder. But you can undestand the words pretty well, even with distorted or noisy speech modulator samples.

 

Other than that I generally believe it is a good idea to follow the concept critical bands in hearing, and then choosing a resolution within each band as needed for transporting more definition into the result. log spacings are probably already close, but you can only get the best possible result by understanding the basic parameters of psychoaccoustics.


Edited by OopsIFly, 07 March 2017 - 16:17.


#9 gentleclockdivider

gentleclockdivider

    Guruh Motha Fakka is Levitating and Knows Everything About Renoise Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • 2476 posts
  • Gender:Male
  • Location:something like a sea or a river...just moving
  • Interests:slacking

Posted 07 March 2017 - 18:29

Wel of course modulating , the envelope follower which acts upon the frequencies of themodulator , is then used to control the amplitude of the corresponing frequency bands or sine waves  of the carrier .

Mind you I used the built in primary ( crappy ) 4bp filters , the new core based zdf filters are way better but consume way more cpu.,

 

Here's another example ;simple drumloop (drumatic 4) ,  converted into  256 sine waves , also playing with  envelope follower decay setting .

Same as screenshot shown above , pretty simple stuff , and good results .

Get reaktor ...you'll Love it :)

https://app.box.com/...qdb1dog8oxhlbce


Edited by gentleclockdivider, 07 March 2017 - 18:33.

THERE IS NO RELIGION SUPERIOR TO THE TRUTH


THERE IS NO TRUTH SUPERIOR TO THE ONE I CHOOSE

C'EST MIEUX D'ETRE BELLE ET REBELLE


QUE MOCHE ET REMOCHE



RUBIO RABIA

 


#10 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 07 March 2017 - 20:08

cool might be tempted to invest, on the day they offer a proper linux version.

 

about anything else...I just have a proud but silly spleen about keeping things renoise native, without plugins other than for generating basic sample material to work with. it is a masochistic quirky crazy set of limitations, and endless fun for me.



#11 gentleclockdivider

gentleclockdivider

    Guruh Motha Fakka is Levitating and Knows Everything About Renoise Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • 2476 posts
  • Gender:Male
  • Location:something like a sea or a river...just moving
  • Interests:slacking

Posted 07 March 2017 - 22:29

I am all for limitations , learn a tool inside out .

That's why I have reaktor ...where renoise comes short , reaktor comes in .

Been on it for almost   8 years , don't need anything else really ....just beware the rabbit hole... 

 

On a side note ..renoise is also capable of some additive stuff ( only limited to 10 voices ) 

 

Just draw a long sample ( few seconds ) , this will function as your partial  envelope  , draw some shapes 

Make  a couple of them in the same instrument (limit is 10 voices ) ) , route them all separate fx lanes , insert ringmods on each effect lane ..tune ringmods to taste 

 

https://app.box.com/...2m2qhcr52d3fr3u


Edited by gentleclockdivider, 07 March 2017 - 22:40.

THERE IS NO RELIGION SUPERIOR TO THE TRUTH


THERE IS NO TRUTH SUPERIOR TO THE ONE I CHOOSE

C'EST MIEUX D'ETRE BELLE ET REBELLE


QUE MOCHE ET REMOCHE



RUBIO RABIA

 


#12 triple zero

triple zero

    Member

  • Normal Members
  • PipPip
  • 52 posts

Posted 30 April 2017 - 15:23

O_o holy crap this one sounds AWESOME!! Is this still based on my old 4-band native vocoder?

(I only have time to check the preview right now, not to inspect the .xrni)

#13 OopsIFly

OopsIFly

    Guruh Member

  • Normal Members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 814 posts
  • Gender:Male
  • Interests:...daydreams... -VS- ...propaganda...

Posted 30 April 2017 - 16:02

Check the xrni, theres quite some design in it to make it sound as nasty as it does. The demo was done with 2 instances of the oopscoder (and a delay effect), only modulated by the very same sample you hear in the beginning of it, though it is chopped a bit and pitched up for the 2nd vocoder.

 

It was made from scratch, but as far as I can remember your vocoder it works in a very similar way, signal followers after filters gaining the matched filtered carrier.

 

I find the secret to making it sound as clear is using very steep filters (butter n8) that will make the sound more defined and less blurred, and also the placement of the frequencies is important.

 

Really, this was rather for fun, I still have the more advanced vocoder on hold, it is waiting for me to read through some accoustics papers because I want to make it sound as neutral as possible, with formant shifting hitting sweet natural spots rather than making it sound like a granny or some daft moron...unless that effect is actually wanted.


  • TheBellows and robohymn like this

#14 robohymn

robohymn

    Super Advanced Member

  • Normal Members
  • PipPipPipPip
  • 138 posts
  • Gender:Male
  • Location:Victoria, BC

Posted 12 May 2017 - 18:07

Bloody awesome, love it. Thanks!!





Also tagged with one or more of these keywords: vocoder, native instrument, robot voice