Formantic modelling

Have you ever thought that some guitar or synth patches sound like a specific vowel (in a phonetic sense, not as written vowels). There is even an effect that is named after the phonetic vowels it sound like: A wah-wah, where speaking its name make you vocalise the transition from [u] to [a] (see/hear IPA vowel chart with audio for reference).

So why don’t we try to approach sound design the other way around. Ask yourself which vowels is most similar to the sound you what to create/recreate (mind that this works best when using waveforms with many harmonic components, like sawtooth waves and distorted guitar signals). When you have picked a vowel, there is only one other thing you need to know about: Formants. As it turns out, the human vocal tract works like a series of resonant filter, with the position of the tongue controlling the resonance frequency of some of these filters. It is the resonance frequency of these filters that makes us perceive a sound like a certain vowel or like something that resembles spoken language. The resonance peaks in the amplitude response or in the resulting signal are called formants. As the table on the linked wikipedia page suggest, only two of these peaks suffice to make a sound similar to the sound of a vowel.

So how can we use this in sound design? All you need is a two-band parametric equaliser. You select the vowel you want your sound to be similar to from a format table or map (again cf. the wikipedia page) and tune the bands of the equaliser accordingly. You may want to use high Q / low bandwidth and significantly boost the bands. To get the most out of it in a guitar effects context, I also suggest turning off speaker modeling, as these models introduce other frequency response maxima in your signal chain. Use a non-resonant/first order low-pass if you want to get rid of high frequency components instead.

If you make the frequencies of one or both of the formants variable, you can also create new versions of the wah-wah effect. But as they don’t sound like ‘wah’ any more they of course need a different name. So maybe you can be the first to popularise the ay-ay or why-why pedal – by using formants for sound design.