6Adding F1: The F0/F1/F2 Processor

F1 → apical electrodes, F2 → basal electrodes

Adding the first formant F1 to F0/F2 gave the vowel its second anchor. The processor lights one apical electrode for F1 and one basal electrode for F2, scaling their pulse heights by the formant amplitudes A1 and A2 and pulsing at the F0 rate. It improved vowel recognition — but it still represents the spectrum with only two moving points, and consonant cues above F2 are missing, which is what drove Multipeak. Schematic.

How much of the spectrum survives?

The feature-extraction bet was that a few perceptually chosen numbers — F0, F1, F2 — carry most of speech, so the rest of the spectrum can be discarded. CIS made the opposite bet: don't decide what matters, just send every band's envelope and let the brain choose. When the feature estimates are correct the formant approach is efficient; when they are wrong, the discarded spectrum cannot be recovered — and that robustness is why the field swung to filter-bank methods. Schematic.

Adding the First Formant: F0F1F2 Improvement Over F0F2

Encoding the first formant alongside F0 and F2 enriched the transmission of voicing, duration, and envelope cues. Open-set sentences without context jumped from 31% to 64% (+33 pp) and spondees-in-noise from 37% to 75% (+38 pp). This step established that more of the spectral envelope, not just a single peak, was needed. Verified means from Tye-Murray 1990 (n=5).

Synthesise a vowel — the two numbers a formant tracker chases

Two resonances do most of the work of a vowel: F1 (lower when the mouth is closed, as in /i/ and /u/) and F2 (higher when the tongue is front, as in /i/ and /e/). Play the vowel and slide F1 and F2 — the percept walks around the vowel space. This is precisely what the F0/F2 and F0/F1/F2 strategies tried to track and transmit; when their estimate of these two numbers was wrong, the wrong vowel was sent. Synthesised in your browser.