Sound Design Workshop: Talkbox Redux

Image placeholder title

The popular talkbox effect from the 1960s imparts a vocal character to audio. It is most often used with guitar and keyboards, but you can use it to process any instrument or vocal. You'll find lots of video examples on YouTube, including one by Stevie Wonder that highlights a talkbox's similarity to a vocoder, and you'll find more examples, a detailed history, and links to hardware models at

The talkbox performs its magic by using your mouth and vocal tract to filter the sound. Antares Throat (Mac/Win; $199 [MSRP]; is an effects plug-in that applies physical modeling of the throat to an audio source. That makes it a natural for producing talkboxlike effects. Throat is not a talkbox emulation, but used in the wrong way — meaning not necessarily as intended by Antares — it yields similar results.

By the Throat

Image placeholder title

FIG. 1: Antares Throat lets you adjust the shape of the vocal tract in five sections from larynx to lips.

Throat divides the vocal tract into five sections between the larynx and the lips. You can adjust the width of each section, and you can adjust the length of each section except the first (see Fig. 1). You can also set global offsets to throat length and width as well as modify the waveform produced by the vocal cords. Making occasional adjustments to the nine individual length and width settings and actively modulating the globals and waveform are the keys to turning Throat from a throat modeler to a dynamic talkbox effect (see Web Clip 1).

Antares makes no provision for remote control of its parameters, so you need to use your plug-in host's tools for automation and MIDI continuous controller mapping. Most DAWs offer both, but you may need to dig under the hood to set up controller mapping.

I like to map the Model Throat section's Length and Width sliders (the global length and width offsets) to the same MIDI footpedal and give them ranges of roughly 0.75 to 1.30 with opposite polarity. Pressing the pedal then increases the width while reducing the length. I map the Model Glottal section's Pulse Width slider to a mod wheel with range 60 to 10 so that pushing the wheel up lowers the pulse width, producing a more nasal sound.

I also map the five section widths to sliders and the four section lengths to knobs on my control surface because graphically the widths are represented vertically and the lengths horizontally. I use these knobs and sliders to create a shape that works well with my source material when I move the footpedal and mod wheel. You can adjust the section length and width settings in real time, although that tends to be less dramatic than changing the global offsets and waveform.

Try using track automation instead of working the footpedal and mod wheel manually. For throaty emulations of phasing and flanging, create a short automation clip and loop it. As an alternative, you might use an application such as Five12 Numerology (, Cycling '74 Max (, or Plogue Bidule ( to create MIDI controller sequences and LFOs to manipulate any or all of the aforementioned parameters.

Mouthing Off

Because Throat analyzes incoming audio for pitch and formant information, which works best on unprocessed, solo material, you'll get the cleanest results from that sort of audio. On the other hand, the talkbox effect is anything but clean, and you can use Throat in this context to process just about anything. Try it with leads, bass lines, rhythm tracks, pads, ambient sounds, and even percussion (see Web Clips 2, Web Clip 3 and 4).

You may need to make one-time adjustments to the other Throat settings to accommodate your source material. If you get high-pitched artifacts, try changing the Precision setting in the Source Throat box. For solo source material such as clean instrumental or vocal leads, match the Vocal Range and Source Glottal Waveform settings to the material. For other source material, adjust these settings by ear — they may or may not have a significant effect. The Add Breathiness sliders introduce highpass-filtered noise, and a little of that goes a long way.

Although I've emphasized motion, don't overlook Throat's value as a stationary effect. Experiment with the individual length and width, the global offsets, and the glottal pulse-width settings. It's tempting to think of the graphic display as an EQ curve, which it is not, but the results have a vocal, resonant-EQ flavor (see Web Clip 5).

Len Sasso is an associate editor of EM. For an earful, visit his Web site