Speech intelligibility

Speech sounds are controlled by the position of lips, teeth, tongue, and velum, which establish a set of acoustic resonant frequencies that characterize the vocal tract. These resonances are excited by harmonic-rich, quasi-periodic pulses of air from the vocal chords for voiced sounds and by the hiss of turbulent air passing through a constriction in the vocal tract for unvoiced sounds. The resultant speech spectrum shows these characteristic resonances or formant frequencies.

The ear performs a short-term spectral analysis of the speech sounds over an interval of some 1/8 sec with a frequency resolution of 50 to 500 Hz, depending on frequency, thus determining the characteristic resonant or formant frequencies of the vocal tract which are deciphered by the brain as a particular speech sound or phoneme.

Intelligibility can be measured either experimentally in terms of syllable, word, or sentence test scores or in terms of a calculable empirical measure known as the articulation index (AI). Test scores can be obtained only after the communication system has been built or simulated in detail. The AI,...