In speech synthesis, formant filters are often used to simulate formant filtering by the vocal tract. Passing a “buzz source” through only two or three formant filters simulate different vowel speech sounds. As a result, speech is fully intelligible through the telephone bandwidth (nominally only 200-3200 Hz). A formant is a resonance in the voice spectrum.
A single formant may thus be modeled by using one biquad (second-order filter section). For example, in the vowel (a) as in “father”, the first 3 formant center-frequencies have been measured near 700,1220 and 2600 Hz, with half-power bandwidths 130, 60 and 170 Hz.
In principle, the formant filter section is in series. As a result, there is no need to specify the gains for the formant resonators, only the center-frequency and bandwidth are necessary to specify each formant, leaving only the overall scale factor unspecified in a cascade (series) formant filter bank.
Emulating the vocal tract
In the musical world, formant filter basically emulates the vocal tract to add vocal quality to synthesizers and other instruments. When your mouth vowels such as “eee eye oh” the resonances of different areas inside your mouth create simultaneous high-Q acoustic band-pass filters. This is one way we recognize people by their voices.
The fundamental pitch of speech depends on the tension of the vocal cords, but just as important is the complex filtering that occurs acoustically inside your mouth. Where a wah-wah filter comprises a single narrow band-pass filter, a formant filter applies 3 or more high-Q that you can tune independently. The relationship between center frequencies and how the amplitude of each frequency band changes over time is what makes each person’s voice unique.
Vocal tuning, or pitch correction, is common in pop music. Sophisticated formant filtering has improved the effect considerably compared to early software plugins. Small changes in the fundamental pitch of the voice can pass unnoticed. Nevertheless, a change large enough to create a harmony parts benefit from applying of formant filtering similar to that of the original voice. This reduces the “chipmunk effect” substantially with the result of a less processed sound.
In formant synthesis, the goal is to simulate the strong resonant structure of many real instruments. Here, one or more formants dominate the spectrum of the output sound. Some analog synthesizers have one simple high-pass filter after the low pass filter. With this, they are giving some additional control over the bandwidth of sounds, and thus a simple type of formant.
Additional Resources & Source Texts