Great public speakers are made, not born. Quantle estimates the number of syllables and words you say, computes the length of pauses you make, estimates your pace, pitch, power and (most importantly) their variability. The app development is inspired by the need to evaluate talk effectiveness without a human coach in a fair, repeatable, and unbiased way. Quantle works for languages in which the number of syllables in a word is determined by the number of vowels, although we tuned Quantle primarily for the English language.
To get the most accurate performance, a smartphone running Quantle should be placed within a few meters from the speaker. Quantle uses an adaptive online algorithm to compute talk estimates. It does not work for multiple speakers, with background music or other disturbing sounds. By default, it neither stores the audio stream nor communicates any of its parts. We used a collection of 1000+ TED talks to tune algorithm performance. Quantle is >82% accurate in estimating the basic metrics on average. For more performance information, please see: https://bit.ly/2Oaa0FK.
Quantle does processing locally in real time and can work in the flight mode. This design has three implications: (1) Quantle does not interfere with the surrounding hardware, (2) it is power-aware, since over 90% of energy is spent to operate the built-in microphone and the screen, and (3) audio data and processing results are not shared with a third party therewith preserving speaker’s privacy.