Apple's Personal Voice lets people newly diagnosed with ALS or other speech-threatening conditions record prompts for about 15 minutes on an iPhone, iPad or Apple-silicon Mac to build an on-device synthetic voice, usable through Live Speech in calls and conversations.
15 minutes
Time to build a Personal Voice
Details
Maturity
Established
Promoter
Apple Inc.
Period
2023–present
Keywords
assistive technology, speech accessibility, big tech, ALS/MND
Context
Personal Voice, shipped with iOS 17 in 2023, helps people at risk of losing their speech - most notably those newly diagnosed with ALS - build an on-device synthetic voice.
Activities
Users read a randomised set of prompts aloud; on-device machine learning builds the synthetic voice in around 15 minutes with no audio uploaded to Apple's servers, and the voice integrates with Live Speech for calls and conversations.
Results
Apple developed the feature in consultation with the ALS community and has published a technical description of the on-device modelling approach, but no independent figures on adoption or communication outcomes have been released.
Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
Apple Machine Learning Research team
Conditions for success
on-device processing preserves privacy
consultation with the ALS community (Team Gleason)
Common failure modes
English only at launch
requires an Apple-silicon device
no independent outcome data published
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Google Research trains speech-recognition models to understand atypical speech (ALS/MND, dysarthria, Down syndrome, stuttering) using samples donated by people with …