Thomas D. Carrell
Commun. Sci. & Disord., Northwestern Univ., Evanston, IL 60208-3570
While signal-to-noise ratios range from about 50--90 dB in laboratory experiments, measurements of speech levels in most natural environments show much poorer listening conditions. For example, Teder [Hear. Instrum. 11, 32--33 (1990)] measured signal-to-noise ratios ranged from a high of 13 dB in a carpeted office to low of 1 dB in a 1986 Chevy Nova traveling at 55 mph. These numbers indicate an immense difference between speech signals presented to listeners in laboratory experiments as opposed to the real-world environment. Although speech intelligibility has been well studied in noise, there has been little study of the acoustic characteristics of speech that allow the message to be separated from the background noise. One characteristic of voiced speech that shows great promise in this regard is amplitude comodulation. In the present research, amplitude comodulated tone-analog sentences [Carrell and Opie, Percept. Psychophys. 52, 437--445 (1989)] were presented with simultaneous white noise and multispeaker babble at 0-, 5-, and 10-dB signal-to-noise ratios. It was found that the beneficial effect of amplitude comodulation was greater at the lower signal-to-noise ratios.