Loud phonation: subglottal pressures and Closed Quotients

Jurjaan Snelleman, Huib Versnel and Philippe Dejonckere

Subglottal pressure is a major factor determining the sound level of speech. In conversational speech, subglottal pressure is around 0.5 kPa. In singing, pressures of about 1 kPa are reported. Professional singers may show pressures up to 10 kPa. However, little is known about subglottal pressures in loud speech in persons without vocal training. This may be due to the invasiveness of direct measurements. In very loud vocalisations intra-oral measurements are not reliable. Oesophageal pressure has been shown to be a useful indirect measure of subglottal pressure if the effects of in- and expiration have been accounted for.
Our subjects were males and females with normal hearing and a normal untrained voice. They were challenged to speak loudly in a natural fashion by presenting background noise varying from 70 to 90 dBA. The subjects read short Dutch sentences, which had to be intelligible for a second subject in the chamber. In addition, they shouted /h/-vowel syllables. The subglottal pressure was measured indirectly with a solid-state catheter with three microtransducers (Unisensor AG), inserted through the nose into the oesophagus. The acoustic signal was recorded with a directional microphone (Shure Beta 54) near the mouth. For the electroglottographic (EGG) signal we used a soft neckband with two monopolar silver electrodes on the skin over both thyroid alae and a custom build amplifier. The signals were simultaneously recorded on pc.
Sound levels of speech and shouting varied over a 30-40 dB range (90-130 dB SPL near the mouth). We found subglottal pressures from 0.3 to 4 kPa during speech to 10 kPa when shouting. A strong correlation was found between subglottal pressure and sound level. We conclude that untrained individuals of either gender can produce very high subglottal pressures in noisy environments, especially when shouting.
The Closed Quotient, derived from the EGG signal with Voce Vista, is supposed to correlate with vocal efficiency / sound level. The individual Closed Quotients will be compared with the other data.

Jurjaan Snelleman, Huib Versnel and Philippe Dejonckere

Affiliation: Dept. of Otorhinolaryngology, University Medical Centre Utrecht, The Netherlands