Theory 20/Speech & Phonetic Science
Intelligibility Is a Modulation Ratio
The claim
Speech intelligibility in noise is set by the ratio of speech envelope modulation depth to masker modulation, so a fluctuating masker can be less damaging than a steady one at the same level.
01The mechanism
Listeners recover speech by tracking envelope dips. A masker that fluctuates independently fills those dips, while a steady masker leaves them open even at higher level.
The ratio of modulations determines how much of the envelope survives, which is why signal-to-noise ratio alone mispredicts.
02What this is not
Signal-to-noise models
Those use level ratios. This uses modulation ratios, which explain the fluctuating-masker effect.
03Predictions
- P1Intelligibility should track the modulation ratio.
- P2Steady maskers should be less damaging at equal level.
04Refutation
Intelligibility tracking SNR at controlled modulation would favour the level model.