Tutorial 3: Vocoder Heaven

Vocoders Set Wild Free by Jouni Airaksinen

Vocoders. Don't we all love them? Atleast when used in a good way. I've been asked several times by different artists how to make vocoders. Usually I've been told that it's hard or they cannot sing well, they don't like their own voice or some other excuse which prevents making vocoders. Another common question is which vocoder to use to get good results. I hope that we now can break some of those myths and recommend few software vocoder plugins.

Note: all audio samples are copyrighted!


Transformers Soundwave

First I break the myth that you need to be good singer to make vocals. You don't. It helps of course to be able to sing on tune, but it's more important to have lyrics which synchronize to the tempo well - and even that can be fixed later with normal audio editor or in your sequencer.

You don't like your own voice? Well, vocoder will solve that problem in the end, but I'd say it's good to be able to listen your voice atleast when syncing the vocal clips to the song, unless you can setup your sequencer to use vocoder in real-time. Setting up the vocoder plugin into real-time use usually requires some audio routing capabilities from your sequencer. I cannot go into details how to do it in Your(tm) sequencer, but the basic idea in most vocoders is to feed two mono audio sources into one audio track while being able to mute the original sources.

In short, you need one stereo audio where the left channel is your voice and right channel is a synthesizer sound. Make sure that the track channels don't bleed the audio between channels - i.e. use hard pan to left and right on the source tracks. Otherwise you will get very unclear vocoder with sort of reverb already in it.

It is also possible to do the procedure above in an audio editor if you can open mono audio sources into one stereo audio. In Wavelab this is done by just opening two mono files from same open dialog, as Wavelab will opens those files into one audio window as stereo. You might need to rename the files so that the left and right channels go in correct order.

Carriers and Airplanes


As we have already went through about how to setup to feed the vocoder with audio, a detailed explanation of that audio is now in order. The audio consist of the Modulator (voice, singing) and the Carrier (synthesizer, vocoder sound). Some vocoders provide internal Carrier when you feed the vocoder midi data, but personally I prefer to have full control so I use my own carrier sound in the example.

Modulator is used as the source for the formant and carrier as the carrier. There's a more detailed description about vocoders in following page. In short the formant component in your voice is what gives your sound shape.

The carrier gives your voice the character to how it actually sounds and the carrier also contains the pitch. The vocoder is used to replace this component in your voice into a synth sound. The carrier in the vocoder means the new carrier sound you want to have.

Airplanes have nothing to do with vocoders, unless you sing something about airplanes .. There's also a page in Wikipedia about vocoders.

Vocoder, Vo-Coder or Voco-Der-Van-Kuy?

There exists quite many vocoders. Some are better than others, but in general it doesn't matter much which vocoder you use. The only difference is the number of bands, filter quality and how much you can adjust the paremeters. Some provide a internal carrier synthesizer also, but that's not really a requirement.

Some hardware synthesizers contain vocoder as effect (you can feed the synthesizers with voice through an audio input) and then there exists also solely hardware vocoders such as Korg DVP-1. DVP-1 is worth of mentioning, because it is the vocoder that Laserdance used in the 80s.

In this article we will concentrate more on software vocoders. The ones I recommend are the Cubase Vocoder (included in Cubase, see also mdaVocoder), Synapse VC-1 Vocoder (included in Orion Platinum as DX effect), Native Instruments Vokator, AKAI DC Vocoder, Fruity Vocoder (included in Fruity Studio), Magix Studio Vocoder and then there is also Prosoniq Orange Vocoder.

In my opinion Synapse VC-1, Vokator, AKAI DC Vocoder and Magix Studio Vocoder are the best vocoders. I haven't used Cubase nor Fruity vocoders so I cannot judge them. There's also the free AnalogX Vocoder, but I wasn't too happy with the results that I got. The sound was choppy and grainy, not to mention that it's not a VST plugin. In our example we use the Synapse VC-1 vocoder.

Preparing the Modulator

Finally we can go into the example I've been talking about. The example is the vocoder track from the song Lights of Destiny on my album. Some of you might have heard older version of the vocoder part already, but we use the new 2005 version now.

The first problem is of course to write the lyrics. Originally I had longer lyrics, but when I was preparing the modulator the longer lyrics didn't fit well, so I left some parts out of the lyrics. Here are the final lyrics:

Modulator in Wavelab's montage view being cut and pasted

Lights of destiny, guiding me,

harbour far away, light house at horizon,

starmap of universe.

Harbour far away, lights of destiny.

starmap of universe. guiding me.

Lights of destiny, of universe.

Light house at horizon.

Lights of destiny.

When recording the vocals, it's best to use headphones to listen the song while singing to the microphone which is being recorded. Also it's good to be able to hear your voice through the headphones as it helps a lot to get the singing in sync with the song. This way you need less work editing the vocals. Needless to say having the melody made for the part helps a lot while singing. You don't have to sing in tune or too well either, just remember to sing the same part several times during one take so you get more material to copy and paste from. The used microphone doesn't matter much, but of course it's more fun if you have a good microphone. I use Sennheizer e835. You can hear the microphone in use with non-vocoded voice i.e. in Macrocosm's Morning in the Dark and Energon's Game of Love.

Our example shows in a good way that you don't need a good voice sample as modulator. I made a small mistake back when I sung the vocals as I had the song playing with higher pitch and speed. Of course my vocals were completely wrong then, but I was lazy and didn't want to sing again so I used a program to slowdown the vocals. This resulted in a very bad audio quality as the program I used wasn't very good. However that doesn't matter. Actually I think it gave the vocoder sound more character.

After you have finished the recording, you need to process the recorded vocals. Cut the best parts out of it, and cut the sentences and possibly words into smaller clips. Trim them and process with slight EQ and compressor. Choose the best parts and load them into your sampler in your sequencer, to be able to sync those clips into longer vocal track. It takes patience and hours of work if you want to make it properly as the syncing needs to be done with millisecond precision. Although, you can always just punch the vocals in and not think too much about it, but the results will also not be too great (you can hear that in several of the spacesynth songs). It's up to you what kind of result that you are looking for. For short sentences and not actually singing, just punching in is a okay and fast way.

I don't want to bore you with the different stages that took me to do the vocals, but here's the final result. Lots of copy and paste, few fade in/outs and you can also notice the bad quality caused by the slow down process. As I said, it's more important to have it properly synced. It took roughly 3 hours to record and edit the vocals. Also make sure that you run the vocals through a gate effect to prevent annoying bleed of the carrier sound in silent parts when processing with vocoder.


Preparing the Carrier

In our example I've used Native Instrument's Pro-53 for the carrier sound. Usually a saw based carrier sound gives a good results, as slightly detuned saw stack. While playing with chords, it gives good amount of lows and highs. Rich sound so to say. Square based sounds usually gives you a very robotic vocoder results and I did use one as a layered vocoder in our example.

The carrier works best when it is played as a continous sound as the modulator voice controls the gain anyway. You can use chords, single note melody or layer different sounds into one carrier. It depends a lot of what you want and what fits to your song. In general carrier sound works well if it has no attack/decay and some release so that the notes are played smoothly. The carrier does sound boring and I've included the Pro-53 preset and midi data so that you can see how it was made.


Vocoder Magic

Synapse VC1

Now it's time to do the vocoder magic. Route both the modulator and the carrier to the vocoder the way I have described earlier. It depends on your program how to do that. Either you can use just plain stereo audio, send fx routing or mixer auxiliary bus routings. In Orion Platinum I use bus routing to route both sounds panned separately into one axiliary bus which then has the vocoder as a insert effect. Output of this bus channel is the vocoder sound. It's even possible to have source audio coming live from microphone and from synthesizer for instant real-time vocoder fun.

I prefer the routing for a couple of reasons. First, you can easily tweak each track separately and you can also EQ both tracks. I usually make the modulator and the carrier more brighter and cut low frequencies from the modulator little bit. The vocoder I use has a few settings for the band gains and the filter settings. The vocoder settings do affect the vocoder sound considerably and while using the Synapse VC-1 Vocoder the Envelope Follow makes the sound more punchy or more smoother as if it was singing. Other settings affect more on the sound brightness. Usually you don't need too much of the low frequencies as you need to cut those out anyway when mixing vocoder into your song. In the screenshot you can see the settings that I used in the example.

At this point it is good to apply some EQ and compression to the vocoder result to cut the unwanted frequencies and even the overall level constant. Constant levels on vocals make them sound more professional - atleast in disco music.

That's about it. Now you have a good sounding vocoder sound. I've also provided the square based robot version.


Mixing Vocoders

Vocoder is a mono sound, so to mix it into the song requires some fx. EQ, compression, exciter, chorus, flanger, delay and reverb. As the example is the vocoder part from my song, I've already done the mixing. To mix vocoders you can quite safely use the same techniques as for mixing plain vocals. Remember to leave room for the vocals in the mix by leaving some instrument layers out of the vocal part. As a thumbrule it's better to compress more and raise gain to good level - and then little bit more. I've heard quite many non-professional mixes where the vocals have been mixed with too low levels. When adding reverb I prefer to have reverb which reverbs the higher frequencies because such reverb is more apparent in vocal material even with lower reverb gains.

Remember that you can also layer the same vocals with different effects and EQ. If you listen carefully you can hear that in the example mix I have used layer of different vocoder sound on top the word "Star" to emphasize that word. In my mixing techniques I like to layer the heavily effected vocoder with less effected and original sounding vocoder layer to bring the vocals more to the front and center. That layer has the purpose of adding clarity as well.

The final mixed result of the example sounds a bit thin on the lower frequecies as I've cut quite a lot out of the lower frequencies. That is pretty normal because you don't notice the lack of lower frequncies when the end result is mixed into the song. In fact you just get a more cleaner and professional mix.

I hope that this tutorial has been helpful!



Discuss more this article in the forum


2005-10-22 21:58:03Written by Jouni Airaksinen