I'm hoping some of you guys might share the techniques you use when vocoding, like what you're using as carriers, pre/post processing you do on input/output signals etc.
Ultimately, I'm hoping someone might have some suggestions on getting that heavily modulated Kraftwerk/Speak and Spell type sound but also, as a long time fan of vocoders, I'm interested in seeing how creative you guys are in it's use too.
Share your vocoder technique?
-
- KVRAF
- 6496 posts since 26 Nov, 2004 from Frederick, MD
My contribution to this month's song competition, Free-For-All contains my first ever use of Native Instrument's Vokator (it shows up prominently but briefly in Part 3), so I am probably not even a qualified amateur 
But I can tell you that once you get the hang of how to use that software, it produces brilliant results.
You mention you want to use a vocoder in its classic way, i.e. voice as the modulator and some other sound as the carrier. That's what I did in my song entry, but it's also the most obvious and most cliché. Back in the old days I had a PAIA vocoder that I built from a kit and my favorite way of using it was thinking of it as a complex active filter. I'd take any sustainable sound from a MIDI synth and program a long sequence of sustained notes going through several octaves of pitch modulation. This I'd use as the modulation track. I would then put a guitar or other audio track in as the carrier signal. The sequence would then sweep the audio signal through several octaves of filtering. Very useful for creating controlled and complex filter sweeps. The resulting effect was not obviously a vocoder, but it was unique and interesting.
However, the way I used Vokator in Free-For-All was using two short vocal samples, one of me saying "Set them free" and another of me saying "Free for all." I kind of sang these in a monotone voice and I had actually done the recording outside of the song, so they were not in tempo with the song. This turned out to be a good thing, as I'll explain later.
For the carrier, I began experimenting with Albino 2 using various presets, tweaking them as I saw fit. I later switched to Absynth 3, which produced a very different result. I'd say that Albino in general produced more classic old-school sounds, while Absynth produced clearer, more noise-laden sounds, which meant that the understandability of the results was better, while the tonality was worse. But I'm sure results will vary depending on the sounds used.
I ended up recording over 30 variations of the two phrases, all of which had something I liked in them. The problem became which ones to choose for the song itself. Since the phrases were not to tempo and I had cut them up into one word snips, I decided to layer 5 different vocoder sounds for each word and use volume automation in Cubase to change how much of each vocoder audio clip was used. This allowed me to accentuate the attack transient of very noise-laden vocoder samples, and to accentuate the vowel portion of each word on the very tonal vocoder samples. All of these were mixed down to stereo and then FX were added in the master mix. Native Instrument's Spektral Delay came in very handy in this regard.
Long answer to a short question!
But I can tell you that once you get the hang of how to use that software, it produces brilliant results.You mention you want to use a vocoder in its classic way, i.e. voice as the modulator and some other sound as the carrier. That's what I did in my song entry, but it's also the most obvious and most cliché. Back in the old days I had a PAIA vocoder that I built from a kit and my favorite way of using it was thinking of it as a complex active filter. I'd take any sustainable sound from a MIDI synth and program a long sequence of sustained notes going through several octaves of pitch modulation. This I'd use as the modulation track. I would then put a guitar or other audio track in as the carrier signal. The sequence would then sweep the audio signal through several octaves of filtering. Very useful for creating controlled and complex filter sweeps. The resulting effect was not obviously a vocoder, but it was unique and interesting.
However, the way I used Vokator in Free-For-All was using two short vocal samples, one of me saying "Set them free" and another of me saying "Free for all." I kind of sang these in a monotone voice and I had actually done the recording outside of the song, so they were not in tempo with the song. This turned out to be a good thing, as I'll explain later.
For the carrier, I began experimenting with Albino 2 using various presets, tweaking them as I saw fit. I later switched to Absynth 3, which produced a very different result. I'd say that Albino in general produced more classic old-school sounds, while Absynth produced clearer, more noise-laden sounds, which meant that the understandability of the results was better, while the tonality was worse. But I'm sure results will vary depending on the sounds used.
I ended up recording over 30 variations of the two phrases, all of which had something I liked in them. The problem became which ones to choose for the song itself. Since the phrases were not to tempo and I had cut them up into one word snips, I decided to layer 5 different vocoder sounds for each word and use volume automation in Cubase to change how much of each vocoder audio clip was used. This allowed me to accentuate the attack transient of very noise-laden vocoder samples, and to accentuate the vowel portion of each word on the very tonal vocoder samples. All of these were mixed down to stereo and then FX were added in the master mix. Native Instrument's Spektral Delay came in very handy in this regard.
Long answer to a short question!
-
- KVRAF
- 8705 posts since 24 May, 2002 from Tutukaka, New Zealand
I don't really do anything that fancy with vocoders. I use the included one with SX (which as far as I know is based on the MDA Talkbox - could be wrong though)
Most of the time I use it via midi rather than use audio as a carrier/modulator (can never remember which is which
)
If I do use audio, then I tend to use PWM based pads, often with quite a bit of white or brown noise added - white noise if I'm doing sibilant vocals, and brown noise for other things.
My most used technique is to simply use a pad pattern that I've already got in a song - then I insert the vocoded vocals occasionally instead of the pads themselves. Most of the time I tend to use vocals more as a rhythmic device or a general feel thing rather than to actually say anything in particular - so they fit in with that approach well.
The other thing I use them for is gated pads - I don't particularly like those VSTi with rhythmic gates on them - they're too rigidly quantised for my tastes, so I simply find a drumloop that feels right and apply it to the vocoder for my pads - that way I can either get it exactly in sync with my main drums, or I can have it different, but with a syncopated feel. And it's also an awful lot easier than setting up a midigate - which is how I used to do that.
As for pre/post FX...I've recently had some good results by applying reverb to vocals and then rendering to audio (quite heavily reverbed) - then I vocode that rather than get a dry vocoded vocal and add reverb later. Sometimes I find that makes the vocal sound more of a whole somehow - it gets very muddy doing it the other way around.
Phasing is another thing I like to put over vocoded vocals - because you can make them very sibilant, phasers have alot to work on - especially something like Mobilohm.
Another one that I like to do is automate the vocoder over a whole phrase - makes it more organic. e.g. I'll ride the dry-through control, so that certain words will come through the vocoding dryer - it emphasises those words in a nice way - like a reverse FX almost - because vocoding is so obvious, any word that is less vocoded comes through like a sledgehammer. And it also makes applying FX like delays to just those emphasised words alot easier. Lots of delay on vocoded vocals can be too noisy, whereas just automate a send to hit that one particular dry-ish word, and it makes a huge impact (to me at least )
Because of all that fine detailed stuff - it can take me ages editing automation - consequently I usually render my vocoded stuff to audio - easier that than copying/pasting automation every time I want to move something.
Another one is to change the number of bands. If I want a vocoded vocal more as a background rhythmic device, I tend to use less number of bands. But I might use exactly the same vocal later in the track as a main musical feature but with more bands to make it more ethereal and instrumental. In which case I'm using it as an instrument in it's own right - I'll replace the pad with that type of vocal.
And I much prefer female vocals with vocoders - don't know why. Even if I do my own vocals and use them completely wet vocoded (which I have to cos I'm a crap singer
). Female vocoded vocals are just so much sexier IMO.
Most of the time I use it via midi rather than use audio as a carrier/modulator (can never remember which is which
)If I do use audio, then I tend to use PWM based pads, often with quite a bit of white or brown noise added - white noise if I'm doing sibilant vocals, and brown noise for other things.
My most used technique is to simply use a pad pattern that I've already got in a song - then I insert the vocoded vocals occasionally instead of the pads themselves. Most of the time I tend to use vocals more as a rhythmic device or a general feel thing rather than to actually say anything in particular - so they fit in with that approach well.
The other thing I use them for is gated pads - I don't particularly like those VSTi with rhythmic gates on them - they're too rigidly quantised for my tastes, so I simply find a drumloop that feels right and apply it to the vocoder for my pads - that way I can either get it exactly in sync with my main drums, or I can have it different, but with a syncopated feel. And it's also an awful lot easier than setting up a midigate - which is how I used to do that.
As for pre/post FX...I've recently had some good results by applying reverb to vocals and then rendering to audio (quite heavily reverbed) - then I vocode that rather than get a dry vocoded vocal and add reverb later. Sometimes I find that makes the vocal sound more of a whole somehow - it gets very muddy doing it the other way around.
Phasing is another thing I like to put over vocoded vocals - because you can make them very sibilant, phasers have alot to work on - especially something like Mobilohm.
Another one that I like to do is automate the vocoder over a whole phrase - makes it more organic. e.g. I'll ride the dry-through control, so that certain words will come through the vocoding dryer - it emphasises those words in a nice way - like a reverse FX almost - because vocoding is so obvious, any word that is less vocoded comes through like a sledgehammer. And it also makes applying FX like delays to just those emphasised words alot easier. Lots of delay on vocoded vocals can be too noisy, whereas just automate a send to hit that one particular dry-ish word, and it makes a huge impact (to me at least
)Because of all that fine detailed stuff - it can take me ages editing automation - consequently I usually render my vocoded stuff to audio - easier that than copying/pasting automation every time I want to move something.
Another one is to change the number of bands. If I want a vocoded vocal more as a background rhythmic device, I tend to use less number of bands. But I might use exactly the same vocal later in the track as a main musical feature but with more bands to make it more ethereal and instrumental. In which case I'm using it as an instrument in it's own right - I'll replace the pad with that type of vocal.
And I much prefer female vocals with vocoders - don't know why. Even if I do my own vocals and use them completely wet vocoded (which I have to cos I'm a crap singer
). Female vocoded vocals are just so much sexier IMO.