Jump to content


Rating

* * * * * 9 Votes

Submitter

SUPPORT TOPIC File Information

  • Submitted: Nov 15 2014 20:51
  • Last Updated: Nov 22 2014 22:18
  • File Size: 2.49MB
  • Downloads: 883
  • License: Creative Commons CC BY

Share this download

Discuss Goto Download Native Renoise Vocoder Instruments 1

vocoder robot

Screenshots

Audio Previews

00:00

Files in this download

WAT WAT WAT?

Yes. Again, maybe this is old hat and somebody has done this before, but given the new amazing R3 sampler instruments with their bloody sweet ass effects chains and just plain brilliant fuck I love this program, etc etc etc... long story short,

WE GOT VOCODER

Please to check these two versions, a 4 band vocoder and a 9 band one. For some reason the 4 band version sounds cooler to me, which is weird because from most vocoder VSTs I prefer the 8 band. I picked 9 bands for mine because the maths was a teensy bit cleaner (because something off-by-one somethingsomething).

In all fairness this is mainly a proof-of-concept, just to show that it can be done (and it's not even that complex, just a bit of copy-paste work). Myself I still prefer to use the great MDA Vocoder VST (which is available for Linux and sounds great), because it's easier to tweak and feels a bit less like a hack ;-) On the other hand, if you ever wondered how the fuck does a vocoder actually work on the inside, get these files, follow the effect chains, and if you have questions do not hesitate to ask!

Basically:

You got the MODULATOR (voice) split into a bunch of frequency ranges using Sends to FX chains with bandpass Filters, which are then fed to a Signal Follower to control the carrier bits that come next. The output of these FX chains is in fact discared (gain -INF dB).

Then you got the CARRIER (low beepy sawtooth, usually), which is also split into the same bunch of frequency ranges using Sends and bandpass Filters. Except these frequency ranges are amplitude modulated using a Gainer that is controlled by the corresponding Signal Followers from the previous step. The output of these FX chains are all mixed together into the final output, producing ROBOT VOICE.

The bandpass frequency ranges are from 300Hz to 3600Hz, which is (roughly) where the most important parts of human voice are located. I limited to this to make best use of the least amount of bandpass filters / sends / instr.FX chains.

There's also a NOISE channel, which in my vocoders is the amount of frequency energy above 3kHz, which is used to amplitude modulate (with another Gainer) a filtered white noise sample. This helps to articulate the consonants in the ROBOT VOICE. This part needs to be tweaked quite a bit more, it sounds a bit shitty currently. There's many tricks to improve it, such as instead of using filtered white noise, using a high pass filtered version of the MODULATOR voice sample (the MDA Vocoder does this, afaik). I want to maybe try a combination of these, using the voice input with a LoFiMat noisifying effect maybe. Long story short, you need this NOISE channel if you want to make the vocoder intelligble, but there's many ways to do it, as long as you somehow mix in something that enhances the original consonants.

It's got a single macro control to adjust the Q (bandwidth/resonance) of the bandpass filter banks. There's much more stuff that could/should be under macro controls, such as the Signal Follower parameters, frequency ranges (using Hydra), the NOISE channel, etc etc etc. I didn't do this because it is proof-of-concept and I kinda believe a vocoder is probably better done with a good VST than native in Renoise (I just wanted to show off this amazing hack, okay).

For the 9 band version, I doubled the bandpass filters to make them "stronger", because without it wasn't even just slightly worse than the nice 4 band robot voice, but pretty much plain shitty.

The demo voice sample that is packed in these instruments is Tyrion from Game of Thrones, it was just such a great clean sample that vocodes beautifully, I used it for this demo, but take care what you use it for because it's obviously copyrighted. The whole point of this instrument is that you can super easy drop-in/replace whatever voice sample you want, anyway.

Finally, I kinda wonder if I should post this in the Downloads section or the Tips/Tricks section. I wouldn't mind having it as a thread in Tips/Tricks if that encourages discussion. I'm sharing this for the feedback, mindshare, discussion, as I know there's a bunch of like-minded DSP/sound-hackers hanging around these forums--not so much because I think this is a useful instrument for your average Renoise-using musician (you're better off using a VST, but if you like this, by all means! have at it! and let me know ;-) ).





  • 494 Total Files
  • 9 Total Categories
  • 128 Total Authors
  • 129143 Total Downloads
  • Jomox XB-09 Latest File
  • danoise Latest Submitter