A talker's fundamental frequency (the pitch of their voice) varies continuously in typical speech. Computer programs now enable modifications to the fundamental frequency which give us a tool to study what role prosody plays in speech understanding. Sharon Miller worked on this project while she was earning an MA degree with an emphasis in Audiology. As part of Sharon's thesis, which is published in Journal of the Acoustical Society of America, we examined the ability of listeners to understand key words in sentences in which she had modified the pattern of the voice pitch. The sentences were presented in background noise. Here are a few examples:
For reference, listen to this normally spoken sentence.
This is an example of monotone or flattened fundamental frequency.
Here is a sentence with variation, but it's unnatural. The fundamental frequency is sinusoidally frequency modulated (FM) with a modulation frequency of 2.5 Hz.
We found that ANYchanges made to the fundamental frequency of a talker's voice made speech understanding in noise worse. Even though the example with FM modulation sounds more natural (at least to some of us) than the flattened example, speech understanding was poorer in that condition than the flattened one. That's why we were surprised when we examined the ability of people to understand Tony's speech in noise. Tony, shown below, communicates with an electrolarynx. In the photo below Tony is using an electrolarynx manufactured by Griffin Laboratories. His device, unlike most that others that produce monotone speech, has a touch-sensitive button that allows its user to vary the fundamental frequency over time. Tony is shown below with his TruTone electrolarynx.
Professor Peter Watson and I recorded Tony producing sentences with his electrolarynx with a varying fundamental frequency and with the same device set to produce a monotone fundamental frequency. Listeners, under controlled experimental conditions, were able to understand Tony's speech with a varying fundamental frequency better than they were when the fundamental frequency was monotone.
Examples of Tony's speech:
Here is a sample with a monotone fundamental frequency frequency.
Here is a sample with time varying fundamental frequency that Tony produced by pressing the "pitch" button on his device.
The 12% better speech understanding in noise obtained with the varying fundamental frequency represented a significant improvement.
A research article describing this work appears in the American Journal of Speech Language Pathology. The goals of this line of work are to understand the role that prosody plays in speech understanding and to improve the quality of speech production for persons with speech disorders.
Here is an example of a native Korean exaggerating voice pitch in a Sentence.
Here is the same Sentence with a flattened voice pitch.
.
.
Comments