Reason Vocoding 101
Look who’s talking now.
Lo and behold: Reason 2.5 is out. The bad news is, it still won’t do your homework or your household chores for you; the good news is, at least it can talk. The BV-512 Digital Vocoder vocoder is a cutting-edge space-age rendition of a hardware Vocoder, with a whopping 512 bands and an equalizer mode.
A brief history of the Vo(co)der
The year was 1928 and Homer W. Dudley of Bell Telephone Laboratories, New Jersey, embarked on his own private Homer’s Odyssey in search of a way to reduce the bandwidth required for telephony, in order to boost transmission capacity. Little did he know that he was way ahead of his time – nowadays, a similar process is used by mobile telephone operators to squeeze more calls into the systems, by way of digital technology like Dynamic Half-Rate Allocation and Adaptive Multi-Rate Codecs. Homer’s idea was to analyze the voice signal, break it down and resynthesize it into a less bandwidth-hungry signal. He called this process “parallel bandpass speech analysis and resynthesis” and conceptualized it through a prototype named The Vocoder (short for “voice coder”). The Vocoder evolved into a more commercially viable design, got renamed to “The Voder” and was unveiled in front of a large audience at the 1939 World Fair. You can listen to a demo of the original 1939 Voder here:
Sadly, the Voder never made the big time, commercially – perhaps the idea of transforming people’s telephone conversations into hollow robotic blabber didn’t sit well with phone company executives – however, a digital version of the Voder had a cameo appearance during World War II, where it served as a part of SIGSALY, a secure voice communication system used for enciphered speech transmissions between Franklin D. Roosevelt and Winston Churchill. The machine remained a closely guarded military secret until the seventies…!
Yup, the digital vocoder existed in 1942 – and now, some sixty years later, it’s finally your turn to play with one. But how does a vocoder work?
It Takes Two to Tango
The very first thing you need to know about the concept of a vocoder, is this: It needs two sound sources to function. One carrier, one modulator. This doesn’t necessarily mean that it requires two separate Reason devices, however – but, more on that later.
A good (albeit flawed) analogy would be to think of the carrier as the raw material, and the modulator as the mold; the Vocoder makes a cast of the modulator and pours the molten carrier into it. Here’s how it works:
Much like a paper that goes through a document shredder, the modulator signal is sliced up in [X] number of bands, each representing one ‘slot’ of the time-varying frequency spectrum. With BV-512, the number [X] can be 4, 8, 16, 32 or 512. The analyzer then looks at the amplitude of each band. By this process, a spectral ‘blueprint’ is extracted. Simultaneously, the carrier signal is analyzed and divided in an equal number of slices. The frequency characteristics of the modulator can now be ‘superimposed’ onto the carrier.
It is therefore preferable that the carrier features a rich spectral content, so that the vocoder can always “find a match” for the modulator. For example, if the carrier sound is a heavily filtered pad sound with nothing going on in the high frequency range, and the modulator is a recording of vocals, you will get a muffled and unintelligible result. This happens because there is nothing in the carrier material that corresponds to the high frequencies required to produce consonants such as “s” and “t”.
Now that we have an idea of how a vocoder works, let’s try our luck with that good old trademark vocoder sound. For this we will need a vocal sample (the modulator) and an analog synth sound (the carrier). The Vocoder itself needs to be in 8- or 16-band mode in order to emulate an old school analog vocoder, since this is in the range of the number of bands these used to have. With the 32- or 512-band setting you tend to lose the ‘robot’ quality and the human quality takes over.
As for the carrier sound, instant results will be obtained by using a raw, unfiltered sawtooth wave – with the emphasis on raw; the most effective carrier sounds for getting vocals across are usually sharp, loud and unrelenting, and sound pretty awful on their own.
A quick way to create such a carrier sound is this:
- Create a Subtractor.
- Using the “Init Patch” settings as a starting point, open up Filter 1 completely by moving the Freq slider to value 127.
- Make the patch more “sluggish” by adding some portamento, some attack time and some release time. This adds an organic touch to the carrier sound, since the human voice A) tends to glide slightly between notes, and B) doesn’t open and close instantly like an envelope in gate mode.
From this basic setup you can go on to add variations: Maybe you want a monophonic carrier. Maybe you want to enable Oscillator 2 and detune it, or use it as a sub oscillator. Maybe you want to use the noise oscillator, which gives the voice a more raspy, whispering character and accentuates the consonants. Maybe you want to add a Scream 4, which really spices things up, either inserted after the carrier or after the vocoder. Experiment!
Here’s a sample setup: classic_voco.rns
If you’re looking for a less machine-like, more human vocoder flavor, you should move up to the 32- or 512-band setting where the details really shine through. The carrier sound needs two elements – a vocal timbre (preferably a sample) and noise. This is perhaps best done with an NN19 or NN-XT, but here we will take our chances with the Malström:
- Create a Malström.
- Use one oscillator for noise: Try the “Pink Noise” graintable. Route it through the filter. Set it to Bandpass mode and keep turning the frequency knob until you get something that sounds like a long “SSSSSS”.
- Use the second oscillator for the vocal timbre. Experiment with the various Voice graintables such as “MaleChoir”.
- Like in the Subtractor example above, add a little portamento, attack and release time.
The Malström setup: humanoid_voco.rns
Keep toying with the different Malström graintables and you’ll find many otherwordly carrier sounds. Or, if you’re a vocalist yourself, sample your own voice producing a steady “aaaaaaaah”, then “ssssssss”, mix these two together and loop the sample. Use this as a carrier in 512-band mode and you can approximate the immensely popular (and by some, fiercely hated) ‘auto-tune’ effect known from Madonna’s “Die Another Day”, Cher’s “Believe” and many others. Investigate!
Noise is fascinating carrier material thanks to its rich spectral content; White noise is all over the frequency spectrum, which means that there will always be something that corresponds with the modulator signal – it lets you “project a shadow” of the modulator. In theory, this means that extending the decay time on the vocoder will result in a reverb-like effect. Not that you’re in dire need of another reverb now that you have RV-7000, but this is a reverb with a twist. Allow us to demonstrate:
- Download the example file reverb_voco.rns, open it in Reason 2.5 and hit Play.
- Examine the setup: The carrier is a Malström playing the “Pink Noise” graintable, the modulator is a Dr.REX playing a drum loop, and the Vocoder is in 512-band mode. Try turning the Attack time up for a bit of pre-delay on the Reverb effect.
- Turn the Decay knob on the BV-512 down to zero and the Dry/Wet knob to 127. Now you will hear the ‘ghost’ of the drum loop projected on the carrier sound (it resembles heavy mp3 or RealAudio compression).
- Now turn the Decay knob back to where it was, around 87, and do the same with the Dry/Wet knob, back to 50/50 (around 64) – or simply reload the song. Now for the fun:
- Experiment with the band setting on the vocoder. Notice how 4 or 8 bands produces a cool “old school Roland analog beatbox” effect.
- Fool around with the Shift knob on the vocoder. Consider that it can be automated and CV controlled, too.
- Enable Mod B on the Malström. It will control the Motion on the noise graintable.
- Browse through the graintables on the Malström and hear what “Thunder reverb” or “TibetanMonks reverb” sounds like. Explore!
- Bonus round: Switch MIDI input to the Vocoder track in the sequencer, and mute the Dr. REX track. Now you can play the Vocoder bands from your keyboard.
Same, Same But Different
Is there a way to use the same device as both carrier and modulator? Certainly. Is there a purpose? Definitely.
1) Using the same sound as carrier and modulator
Needless to say, if the carrier and the modulator signals are identical, not much will happen. A sound imposing its characteristics on itself will result in status quo. But once you throw effects into the melting pot, things get interesting. Par example:
- Create an NN19, a BV-512, a Spider Audio and a Scream 4. Load a vocal sample into the NN19.
- Using the Spider, split the NN19 output and route one signal to the modulator input on the BV-512, and the other through the Scream 4 and into the carrier input on the BV-512.
- Play the NN19 from your keyboard and mess around with the Shift knob on the BV-512 for interesting results; the Shift parameter now controls the formants.
- (example file: shift_voco.rns)
2) Using the same device as carrier and modulator
Since the NN-XT has multiple outputs, it can be used as both carrier and modulator at the same time. In other words, It will need only one MIDI input source to work. Par ejemplo:
- Create a BV-512. Create an NN-XT and load a strings patch.
- Add one sample Zone and span it across the entire keyboard. Make a separate Group of it and route it to Output pair 3+4. Load a rhytmic sample into this Zone.
- Set the Group polyphony parameter to 1 (monophonic).
- Change the Pitch kbd tracking to zero (fixed pitch) for this Zone.
- Connect the 1+2 outputs to the carrier input of a BV-512. Connect output 3 to the modulator input.
- Now you can play chords on the keyboard and hear the rhythm sample affect the timbre of the sound. (example file: lazy_voco.rns)
You can of course also use two polyphonic NN-XT sounds in the above setup to “morph” any two sounds together: Strings and piano, effect sound and pad… it’s your call.
Hopefully, this exercise has begun to open your mind to the fact that a vocoder is much more than a talking synth thingamabob. A vocoder is a set of bandpass filters, and it imposes the characteristics of one audio signal onto another audio signal. Either audio source can be any sound, and vocals is just one source material out of billions. Now get to work!