Vocal synthesizers blur the line between natural and artificial like no other type of effect.
Since the early days of sound synthesis, engineers have looked for convincing methods to emulate the sound of a human voice.
And while some approaches were more successful than others, vocal synthesis technologies have become a part of the language of modern music production
But what exactly is vocal synthesis? What are the key types, and where are they most useful in a mix?
In this article I’ll break down the most common approaches to vocal synthesis and suggest three creative ways to use them in your workflow.
Let’s get started.
What is vocal synthesis?
Vocal synthesis refers to a group of technologies that simulate the sound of human voices or augment vocal recordings with synthetic properties.
Vocal synthesis began in the 1970s with the introduction of analog vocoders. Short for ‘voice encoders,’ these devices became popular in early electronic music production.
Since then, vocal synthesis has developed with new techniques and technologies, sparking trends in vocal production across musical genres.
Here are the main four types of vocal synthesizer in common use today:
Vocoder
As I mentioned above, analog vocoders were the first widely used vocal synthesis tools.
They work by imparting the characteristics of a speaker’s voice onto a synthesized tone.
A vocal recording or microphone input provides the modulator signal and a synthesizer oscillator provides the carrier signal. The modulator’s amplitude and frequency characteristics are used as the modulation source for the synth's envelope and filter, creating a distinctive ‘talking synth’ effect.
There are plenty of great vintage vocoder plugins out there if you’re looking for the classic version of this effect, including the free TAL Vocoder.
Formant synthesis
As synthesis and acoustics matured, researchers identified unique characteristics of human speech known as formants.
Formants are fixed resonant frequencies that arise from the shape of the singer’s vocal tract. They’re unique to every person and play a significant role in how we differentiate between voices.
Whether using physical modeling or frequency modulation, formant synthesis recreates known formant patterns such as those of vowel sounds to simulate speech.
The Yamaha FS1R rackmount synth was one of the most advanced hardware units to include formant synthesis.
Vocal Resynthesis
Vocal resynthesis is another method of synthesizing vocal timbres that starts with an existing vocal recording.
With this technique, the frequency information in the original audio is analyzed, deconstructed and then remapped onto a new signal.
Unlike a traditional vocoder, the analysis phase extracts information about the harmonic partials that are present in the original vocal.
This gives vocal resynthesis tools even greater control of the resulting timbre of the sound, which is vital for creating a realistic voice.
Our Humanoid plugin uses a complex resynthesis process to dramatically alter the timbral qualities of the incoming vocal.
In addition to hard tuning and pitch correction, Humanoid morphs your voice into a synth waveform as you turn up the Transform dial.
Watch Alex demonstrate how Humanoid combines powerful tuning, formant shifting and resynthesis to create a uniquely synthetic vocal texture.
AI Voice Generators
The latest developments in vocal manipulation use neural networks to map the characteristics of one voice onto another.
By training on many recordings of one speaker or vocalist, AI systems can identify the unique qualities present in their voice.
AI vocal transformation is not a traditional form of synthesis, but a completely new process that uses machine learning in place of signal processing techniques.
This method has some unique advantages, as the neural network can identify subtle qualities in the target signal that can’t be easily modeled in a virtual signal path.
In fact, our TAIP plugin uses this approach to create an accurate simulation of a vintage tape machine.
In the case of AI vocals, the realistic accuracy of the simulated voice is the subject of some controversy, as it raises concerns about the potential for misuse.
Even so, AI vocals have exciting creative applications that artists like Holly Herndon have embraced to explore the boundaries of art and technology.
How to use vocal synthesizers in your mix
With the background out of the way, here are some creative techniques to consider using vocal synthesis.
To take advantage of any of these ideas, you’ll need a capable vocal synthesis plugin in your arsenal.
We built Humanoid to give producers access to multiple vocal manipulation and resynthesis tools with a simple and creative workflow.
If you’re looking for a single plugin that can transform your vocal sound into something new and synthetic, Humanoid is a powerful option for all the techniques I’ll describe in the next section.
On to the tips.
1. Classic robot voices
Robotic voices kicked off listeners’ fascination with synthetic vocals in the early era of analog synthesis.
The uncanny sound is still just as captivating when used effectively.
Humanoid can evoke the familiar sound of retro vocoding, but it can also push the technique into the future.
With the Transform control turned up, the original vocal texture gets forcibly mapped onto a synthesis waveform. Try using a simple square or saw wave to emulate the primitive texture of early analog vocoders.
Or try playing rich polyphonic chords in MIDI mode for a modern, detailed vocoder effect.
2. Augmented natural voices
Modern vocal synthesis often serves to enhance a natural vocal, or exaggerate it’s qualities for a stylish effect.
While Humanoid isn’t exactly subtle, it can work with the incoming vocal texture to create a distinctive synthetic layer that adds interest to an arrangement.
Use Humanoid’s mix controls to balance the blend of tuning, resynthesis and dry vocal to create a robotic hybrid that stands out in the mix.
3. Vocals as an instrument
One interesting aspect of extreme vocal synthesis is that it can completely replace the incoming vocal with a synthetic tone.
This is useful when you want the phrasing and expression of a vocal line without an identifiable voice quality.
With the Transform dial fully clockwise, Humanoid eliminates any trace of the original voice, leaving only the resynthesized synth tone controlled by the vocal input.
Try improvising melodies with Transform at 100% to come up with unique hooks you might not discover on your instrument.
Synthetic voice machine
The singer is often the focal point of a song’s drama and emotional intensity.
Vocal synthesizers give you the ability to alter their sonic persona in ways that can be surprising and uncanny.
If you’re looking for a way to add a unique dimension to your vocal tracks, get back to your DAW and try out the techniques covered in the article.