Introduction
For most people, voice cloning is a productivity tool. For people facing conditions that affect their speech — ALS, throat cancer, Parkinson's disease, stroke recovery — it is something far more personal. It is the ability to keep speaking in their own voice even after their biological voice is gone.
This guide covers how voice preservation works, which tools are best suited for accessibility, and how to get started while time allows.
Why Voice Preservation Matters
Losing your voice is not just losing the ability to communicate. It is losing a core part of your identity. Your voice carries your personality, your accent, your warmth, your humor. Synthetic alternatives — the robotic voices of traditional AAC (Augmentative and Alternative Communication) devices — are functional but impersonal.
Stephen Hawking communicated through a machine for decades, but it was not his voice. Today, AI voice cloning means that is no longer necessary.
A person diagnosed with ALS can record their voice today and continue "speaking" in that voice through a text-to-speech system for as long as needed.
How It Works for Voice Preservation
The process is the same as standard voice cloning, but with heightened urgency and specific considerations:
- Record while you can. The best time to record is when speech is still clear and natural. For progressive conditions, this means as soon as possible after diagnosis.
- Create a comprehensive sample. For preservation, more audio is better. Aim for 30-60 minutes covering different emotions, topics, and speaking styles.
- Clone with a reliable platform. Choose a platform that will store and maintain your voice model long-term.
- Integrate with AAC software. The cloned voice needs to work with the assistive technology the person will use (tablet, eye-tracking system, switch-based input).
Recommended Tools for Voice Preservation
ElevenLabs
ElevenLabs Professional Voice Cloning produces the highest quality clones. Their API integrates with third-party AAC applications. Cost: $99/month for Professional Cloning.
Acapela My-Own-Voice
Specifically designed for voice preservation. Acapela partners with hospitals and ALS clinics to help patients bank their voices. The process is guided and optimized for people whose speech may already be partially affected.
ModelTalker
Developed by Nemours Children's Hospital, ModelTalker is a free voice banking system designed for people with ALS and other degenerative conditions. It is specifically optimized for the medical use case.
VocaliD
VocaliD creates personalized synthetic voices by blending the user's remaining vocal characteristics with a matched donor voice. This is particularly valuable when the person's speech is already partially deteriorated.
Recording Guide for Voice Preservation
Recording for preservation differs from recording for content creation:
Priority: naturalness over quality. A warm, natural recording in a living room is more valuable than a stiff recording in a studio. The goal is to capture the person's authentic voice, including the small imperfections that make it theirs.
Include emotional range. Record happy moments, serious discussions, laughter, surprise. The AI needs these examples to reproduce emotional variation.
Record conversations, not just reading. Scripted reading captures pronunciation but misses conversational patterns. Record natural conversations alongside prepared text.
Practical recording tips:
- Use a quiet room but do not obsess over studio conditions
- A laptop microphone or phone is acceptable if nothing better is available
- Record in multiple sessions if fatigue is a factor
- Include the person's frequently used phrases and expressions
- Record interactions with family members (captures their social speaking style)
Organizations That Help
Several organizations provide free or subsidized voice banking:
- Team Gleason — Provides technology (including voice banking) free to people with ALS in the US
- ALS Association — Funds voice banking programs and connects patients with resources
- My-Own-Voice by Acapela — Offers guided voice banking, sometimes subsidized through healthcare partnerships
- Communication Matters (UK) — Provides AAC resources and voice banking guidance
Integration with Assistive Technology
Once the voice is cloned, it needs to work with the person's communication system:
Eye-tracking systems (Tobii Dynavox, Grid 3) allow users to select words and phrases on screen. The cloned voice speaks them aloud.
Switch-based input for people with limited mobility. Single switch scanning through word prediction, with the cloned voice as output.
Tablet-based AAC (Proloquo, TouchChat) with custom voice integration. Type or select symbols, hear your own voice.
The technical integration varies by platform. ElevenLabs API is the most flexible for custom integration. Acapela provides pre-built AAC partnerships.
A Note on Timing
If you or someone you know has been diagnosed with a condition that may affect speech, start the voice banking process as soon as possible. The quality of the clone depends entirely on the quality of the input. Clear speech produces the best results.
Do not wait for speech to deteriorate. A voice banked today preserves the voice at its best.
Frequently Asked Questions
How much does voice banking cost?
ModelTalker is free. Acapela My-Own-Voice costs vary (often subsidized through healthcare). ElevenLabs Professional Cloning is $99/month. Team Gleason provides technology free for ALS patients.
Can I bank my voice if my speech is already affected?
Yes, though results are better with clear speech. VocaliD specializes in working with partially deteriorated speech by blending with donor voices. Start as soon as possible.
How long does the cloned voice last?
Indefinitely on most platforms. ElevenLabs stores voice models as long as your account is active. For maximum security, use a self-hosted solution like Coqui TTS to keep the model on your own hardware.
Does insurance cover voice banking?
Some insurance plans cover AAC devices and related services. Voice banking itself is increasingly recognized as part of AAC. Check with your provider and ask your speech-language pathologist.
For the technical cloning process, see our voice cloning tutorial. For tool comparisons, read best voice cloning tools.