Faster Text-to-Speeches

Blind people rely mostly on the auditory feedback of screen readers to consume digital information. Still, how fast can information be processed remains a major problem. The use of faster speech rates is one of the main techniques to speed-up the consumption of digital information. Moreover, recent experiments have suggested the use of concurrent speech as a valid alternative when scanning for relevant information. In this paper, we present an experiment with 30 visually impaired participants, where we compare the use of faster speech rates against the use of concurrent speech. Moreover, we combine these two approaches by gradually increasing the speech rate with one, two and three voices. Results show that concurrent voices with speech rates slightly faster than the default rate, enable a significantly faster scanning for relevant content, while maintaining its comprehension. In contrast, to keep-up with concurrent speech timings, One-Voice requires larger speech rate increments, which cause a considerable loss in performance. Overall, results suggest that the best compromise between efficiency and the ability to understand each sentence is the use of Two-Voices with a rate of 1.75*default-rate (approximately 278 WPM).