Voice banking: how to save your own voice
Voice banking means recording your voice while you can still speak clearly, so that a synthetic copy of it can speak for you later through a phone, a computer or an AAC device. People facing ALS, MND, throat cancer surgery or other conditions that affect speech often do it, but anyone can.
This guide compares three ways to do it, from what each provider says about its own service. Start as early as you can, and talk to your speech and language therapist, who can help you choose and set it up.
Why start early
A banked voice can only sound as good as the recordings it is made from. Record while your voice is still the one you want to keep, in a quiet room, with the same microphone each time, and keep a backup of the raw recordings in a safe place: they are worth keeping even if the tools change.
Apple Personal Voice
On iPhone, iPad and Mac, Personal Voice makes a synthesized voice that sounds like you. You find it in Settings, Accessibility, Personal Voice, and read a series of sentences aloud; you can stop and resume later, and the speech is processed on the device. Apple lists iOS or iPadOS 17 or later, iPhone 12 or later and Macs with Apple silicon, in supported languages. The voice works with Live Speech, Read & Speak, VoiceOver and AAC apps that support it.
Acapela My-Own-Voice
Acapela’s My-Own-Voice service is made for people at risk of losing the ability to speak. According to Acapela, it needs only about 10 minutes of recordings and is available in 26 languages, and Acapela lists AAC makers such as Tobii Dynavox and Saltillo among its partners; check its pricing and how the voice reaches your device before you start.
A voice trained from a short recording
In the SuperClonica studio you can train a voice from a single clean recording of 5 to 10 seconds, on a computer with an NVIDIA graphics card, then type any text and hear it in that voice with the free engine, offline: superclonica --text "I love you all. Thank you for being here." --voice my-voice.json --lang en --output message.wav. A short sample captures the colour of the voice rather than every habit of how you speak, so treat it as a complement to full voice banking, for example for recorded messages to family, and keep your longer recordings too.
Use your voice fairly and safely
Your voice is yours: SuperClonica trains voices only from recordings you have the right to use, and a family member can help with the setup but should not make a voice of you without your consent. A banked voice is a communication aid, not a medical treatment; your care team remains the right place for advice on speech and swallowing.