An assessment of fluency enhancement techniques for a telephone device for stutterers

Telephone use is one of the most stressful communication situations for stutterers. We investigated a device to modify stuttered speech spoken into a telephone with the goal of ameliorating the stress and providing greater fluency. The device uses signal processing techniques to detect and correct certain types of dysfluencies. To assess dysfluent telephone input, stuttered speech exhibiting repetitions, prolongations, and blocks were recorded and then processed using phonetic classification technology to detect certain types of dysfluencies, and timescale modification to correct them. In a series of experiments, listeners assessed the quality and intelligibility of the dysfluent (unprocessed) speech vs. the fluency-enhanced (processed) speech. Listeners assessed the processed speech as both more acceptable and more intelligible than the unprocessed speech.