Achieving Natural Fluency using a Metronome based approach

Abstract

This study investigates person's ability to entrain to a stressed metronome to see if this produces fluent and natural speech. The concept of a stressed metronome is unexplored in current research. The literature shows that speaking one syllable on and syllable off to a metronome results in highly fluent, however also highly unnatural speech. This study aims to examine stressed syllable entrainment at three different speech rates to look at the effect this has on the speech naturalness ratings. Three different metronome rates in beats per minute were established to compare the effect they have on speech naturalness. A baseline speed was determined from the average of ten participant's speech rates. The fast and slower rates were determined from a method used in a previous study Twenty participants were recorded reading the stressed syllables highlighted in a passage to the metronome beat. The audio files were saved on a computer and analysed by studying the acoustic and provided waveform displays of the speech using PRAAT. Overall speech rate of each recording and the distance between onset of stressed vowels were analysed to determine entrainment at a gross and precise level. The overall speech rate was determined from where the speech signal began and ended. The distance between vowels were annotated and specific intervals in the speech data were marked. The recordings were then listened to and rated by naïve listeners on a 5 point scale. The Results show that overall, speakers can entrain to a stressed-timed metronome, on both a precise level and gross level. However, this does not increase naturalness of speech in all conditions. To conclude, a possible combination of syllable-timed and stressed-timed entrainment could be an effective method. This is the first step in a process to devising a method to achieve naturalness of speech and fluency simultaneously.