Saylists
An algorithm designed to transform speech therapy by identifying beneficial sounds in songs to enhance repetition with music.
Tech Stack
Overview
Saylists is an innovative algorithm designed to transform speech therapy by incorporating music into repetitive exercises. Developed for Warner Music & Apple Music, this project aimed to make speech practice more engaging and effective through the power of sound repetition. The algorithm identifies and scores specific speech sounds in songs’ lyrics (e.g., S, T, or double letters like in “success”). By calculating the frequency of these sounds, their proximity to one another, and their weight based on the inverse-square law, the algorithm tailors songs for speech therapy exercises.
The algorithm was built to be flexible. We could tune the scoring for sounds that happened at the start of a word, the middle, or the end.
The criteria for this was provided by professional speech therapists. Once we had our songs scored we brought the findings back to the speech therapists who could verify the results.
Screenshots
Key Features
- 1 Detects specific speech sounds in song lyrics (e.g., S, T, double letters) and scores their frequency and proximity.
- 2 Uses the inverse-square law to weight sounds based on their distance from one another, prioritizing closer sounds for more effective therapy.
- 3 Built to scale horizontally, ensuring efficient handling of expanding song libraries and growing user needs.
Technical Challenges
-
The initial version of the algorithm needed multiple tweaks to refine which speech sounds were most beneficial. It underwent numerous iterations to perfect the scoring and sound selection process.
-
Managing over 500,000 songs in the database was a significant challenge, particularly in terms of data processing speed.
Related Posts
Building the Saylists Algorithm
23 March 2026How we scored half a million songs for speech therapy using phonetics, the inverse-square law, and a lot of iteration