Voice Banking
Save your voice for future communication
Review status: Waiting for team review. An AI editorial check was completed on 2026-08-14 by claude-opus-5. It checks clarity, attribution and scope against the sources, and it is not a Compass team or clinical sign-off. Waiting for clinical review. Last updated 2026-08-14.
People who may help
Not everyone needs all of these, and who does a particular job differs between countries and services.
Items
Add an item
Saved on this device only. Never sent to us.
Simple ways to get started with this card. Use these as a checklist, add useful items to My Plan, or create your own.
Planning 3
- Check whether now is the right timeVoice banking usually works best while speech is still reasonably clear.Why it matters: Timing genuinely matters here, because the recording depends on the voice you have now.
- Compare voice banking optionsReview the available services, apps, and any recommendations from your speech team.Why it matters: The options differ in effort, cost, and the result, so a short comparison saves redoing it.
- Schedule your first recording sessionSet aside quiet time and a suitable device or microphone.Why it matters: Recording is tiring, so booking it as real time rather than fitting it in makes finishing far more likely.
Your feedback on these suggested items
Are these useful starting items for Voice Banking? Reference an item by its title in your comment.
#
ALS/MND can affect speech for some people, and a few options exist to preserve your own voice and words in case that happens. They are simple in idea and can mean a lot.
- Voice banking records samples of your speech and uses them to create a synthetic version of your own voice. If you later use a communication device that "speaks" for you, it can sound recognisably like you rather than a generic computer voice.
- Message banking records specific phrases, sayings, sounds or messages in your real voice. That could be the way you say "I love you", a familiar joke, a bedtime story, or everyday phrases you use a lot. These recordings can be kept and used later, and some can be built into communication devices.
People choose to do this for different reasons: to keep sounding like themselves, to preserve phrases that matter to family, or simply to have the option open. There is no obligation. Some people bank their voice and never need it. Others find it deeply worthwhile.
The most important thing to know is that these work best when your speech is still clear, so it is worth understanding early rather than later. The other questions on this card cover the timing, how it works, the different approaches, and what to do if your voice has already started to change.
Explained: what this word meansEvidence for this answer
The sources the Compass team used to write this answer. Highlighted phrases in the answer above correspond to the statements below.
Statement 1 of 2. A voice built from your own recordings can sound recognisably like you on a communication device, rather than like a generic computer voice.
Qualifies this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body puts the limit more strongly: a personalised synthetic voice can never be an exact replica, and it says some people are disappointed and feel the voice does not sound like them.
“It is also worth noting that the PSV can never be an exact replica of a person's voice and therefore some people may be disappointed and feel that the voice does not sound like them.”
Role of the speech and language therapistLink checked August 2026
Statement 2 of 2. Voice banking and message banking work best while speech is still clear, which is why they are worth understanding early.
Supports this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body gives the same timing guidance for both approaches, and puts the point to act before speech quality has changed noticeably. It also says people should be given enough information as early as possible to decide for themselves, which is the reason for raising it early.
“Voice banking and message banking are best carried out before there is significant change to speech quality.”
Role of the speech and language therapistLink checked August 2026
Used across the whole answer
#
The short answer is: sooner rather than later, while your speech is still clear.
Voice banking and message banking capture the quality of your voice as it is when you record. If speech has already started to slur or weaken, the recordings will reflect that, and so will any voice built from them. Doing it earlier generally gives a better, more natural result.
This is one of the things people often wish they had known about early, because it is easy to put off and then find the window has narrowed. That is why it tends to be raised soon after diagnosis, even if speech is completely normal.
At the same time, there is no pressure and no single deadline. Not everyone wants to do it, and that is fine. If you are interested, the practical message is simply: it is worth understanding and, if you decide to, starting while your voice sounds like you, rather than waiting to see what happens.
If speech is already changing, do not assume it is too late. There are still options, covered in a separate question on this card.
A speech pathologist can advise on the best timing and approach for you, and help you get started. If voice banking matters to you, it is reasonable to raise it early rather than waiting to be offered it.
Explained: what this word meansEvidence for this answer
The sources the Compass team used to write this answer. Highlighted phrases in the answer above correspond to the statements below.
Statement 1 of 2. The usual advice is to bank a voice sooner rather than later, while speech is still clear.
Supports this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body places the point to act before speech quality has changed noticeably, and gives the same guidance for message banking.
“Voice banking and message banking are best carried out before there is significant change to speech quality.”
Role of the speech and language therapistLink checked August 2026
Statement 2 of 2. A synthetic voice reproduces the speech as it sounded on the day it was recorded, including slurring or weakness. Standard voice banking does not correct it.
Supports this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
States plainly that banking a voice that is already slurred or weak produces a synthetic voice that is slurred or weak in the same way, and that nothing in the process fixes it.
“Voice banking with dysarthria or dysphonia will result in a PSV with dysarthria or dysphonia – there will be no 'correction' of any impairment.”
Role of the speech and language therapistLink checked August 2026
#
The basic idea is straightforward: you record your voice reading set material, and that recording is used to create either a synthetic version of your voice (voice banking) or a library of phrases in your real voice (message banking).
In practice it usually involves:
- A quiet space and a decent microphone. Background noise matters, so a calm room and a reasonable headset or microphone help the quality.
- Reading a set of sentences or phrases. For voice banking, you typically record a number of prompts. It can be a few hundred, so the system has enough of your speech to work with. How much is needed varies a lot between services, and some newer ones ask for much less, sometimes only a few minutes of speech. For message banking, you record the specific phrases you choose.
- Breaking it into sessions. You do not have to do it all at once. Many people record over several sittings, which also helps if speaking gets tiring. It is usually best to keep the sessions fairly close together, because your voice can change over time and services generally want the recordings to sound consistent.
- Saving and processing. The recordings are stored, and (for voice banking) processed into a synthetic voice you can use later on a communication device.
The amount of recording, and exactly how it works, depends on the service or tool you use. A speech pathologist can point you to options suitable for you and help you get set up.
It can feel like an odd or emotional thing to do, especially early on. Many people find it easier with someone alongside them, and find it becomes meaningful. It is a recording of your own voice and words, kept for whenever it might be wanted. There is no need to make it perfect; the goal is simply to capture your voice while it sounds like you.
Explained: what this word meansEvidence for this answer
The sources the Compass team used to write this answer. Highlighted phrases in the answer above correspond to the statements below.
Statement 1 of 3. How much recording is needed varies a lot between services, and can run to a few hundred prompts.
Qualifies this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body gives a typical range that starts at a few hundred phrases but reaches well over a thousand, so a few hundred is the low end of what it describes.
“To create a PSV, a set-list of phrases needs to be recorded – often between 350 and 1600.”
What is voice banking?Link checked August 2026
Statement 2 of 3. Background noise and the microphone you use affect the quality of the recordings, and so the quality of the voice built from them.
Supports this. Voice Banking Guide — University of Iowa Speech, Language, and Hearing Clinic · Practical guide
A university speech clinic guide covers both halves of the setup. It lists practical ways to cut out noise, such as closing windows and turning off fans, and it says the quality of the audio affects the quality of the voice, which is why it tells people to use an external microphone rather than the one built into a computer.
“Make sure you avoid background noise that can decrease the quality of the recording.”
Voice Banking, Before you BeginLink checked August 2026
Statement 3 of 3. Recording does not have to be done in one sitting, and taking breaks is normal advice rather than a compromise.
Supports this. Voice Banking Guide — University of Iowa Speech, Language, and Hearing Clinic · Practical guide
Its walkthrough for one of the longer services puts the recording at six to seven hours in total, and tells people to take breaks rather than try to finish it in a single day.
“Note: Recording will take around 6-7 hours so make sure you take breaks and do not aim to record in one day.”
ModelTalker, Recording the main sentencesLink checked August 2026
Used across the whole answer
#
These terms overlap and are easy to mix up. Here is the difference in plain terms.
-
Message banking keeps your real voice. You record specific phrases, sayings and sounds exactly as you say them, and those recordings are saved and can be played back or built into a device. It captures how you really sound saying those particular things, including emotion and personality. That only applies to the phrases you record.
-
Voice banking creates a synthetic version of your voice. You record a set of material, and a synthetic voice is built from it that can then say anything you type later on a communication device. It is not limited to pre-recorded phrases. It sounds like you, though, being computer-generated, it may not carry every nuance of your real voice.
-
Voice cloning is a newer approach that also builds a synthetic voice, but using AI that can often work from less recorded material than older voice-banking methods. The quality varies and the technology is evolving quickly.
Many people combine message banking (for treasured, real-voice phrases) with voice banking or cloning (for everyday flexible communication). They serve different purposes.
The right mix depends on what matters to you, how much recording you are able to do, and what is available. The technology, especially cloning, is changing fast, so it is worth getting current advice. A speech pathologist can explain the up-to-date options and help you choose an approach that fits.
Explained: what this word meansEvidence for this answer
The sources the Compass team used to write this answer. Highlighted phrases in the answer above correspond to the statements below.
Statement 1 of 3. Message banking can only play back the phrases you actually recorded. It cannot build new sentences you did not record.
Supports this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body sets out the same limit, and adds the consequence: these recordings cannot be used to build new sentences by typing or spelling them.
“The individual message banking recordings can be added to an AAC device, though playback is limited to only what has been recorded — they cannot be used for generating novel messages by spelling (Costello, 2020).”
What is message banking and how is it different to voice banking?Link checked August 2026
Statement 2 of 3. AI voice cloning can usually build a voice from much less recorded material than older voice banking methods need.
Supports this. ALS Voice Banking and Cloning — Your ALS Guide · Practical guide
The same page that puts older voice banking at 50 to 200 sentences says a cloned voice can be made from about a minute of conversational speech, though more recording gives a closer resemblance.
“With just a one-minute audio file of you speaking in a conversational style, you can get a quality instant voice clone.”
AI Voice CloningLink checked August 2026
Supports this. Communication Devices and Mounts — Guide to ALS · Practical guide
Draws the same contrast from the point of view of someone who could not finish the older style of recording, describing the AI versions as avoiding the large volume of recording.
“There are now AI (artificial intelligence) versions of voice banking that give more realistic speech without having to record a huge number of words and sentences.”
Voice BankingLink checked August 2026
Statement 3 of 3. Message banking and voice banking are often done together, because each covers something the other does not.
Supports this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body treats the two as complementary rather than alternatives, and says both sets of recordings can normally be used on the same device.
“Given the short time it may take to complete both, message banking and voice banking are frequently completed in parallel.”
What is message banking and how is it different to voice banking?Link checked August 2026
Used across the whole answer
#
If your speech has already started to change, it is natural to feel you may have missed the window. That is not necessarily the case, and it is well worth asking rather than assuming.
A few things can still help:
- It may still be worth recording. Even if your voice is not as clear as it was, message banking of important phrases, or banking what you can, may still capture something meaningful. A speech pathologist can advise whether it is worthwhile for you.
- Existing recordings can be useful. Old voicemails, videos, voice notes or recordings of you speaking from before your diagnosis can sometimes be used, especially with newer voice-cloning tools that need less material. It is worth gathering any you have.
- A donor or ready-made voice. If banking your own voice is no longer possible, communication devices come with ready-made voices you can choose from. Some services can also build a voice using a donor, often a friend or family member whose accent and tone are similar to yours, and some can blend more than one donor voice. A speech pathologist can tell you what is available where you are.
So even if the timing is not ideal, it is still very much achievable to have a way to communicate that feels like you. The communication-technology card covers the devices these voices are used on.
The best next step is to talk to a speech pathologist sooner rather than later, since some options become harder over time. They can look at what is possible for you now and help you act on it.
Evidence for this answer
The sources the Compass team used to write this answer. Highlighted phrases in the answer above correspond to the statements below.
Statement 1 of 2. Recording can still be worth doing after speech has started to change, though what it produces will not be the voice as it was.
Qualifies this. Voice banking (clinical information) — Royal College of Speech and Language Therapists · Practical guide
The speech and language therapy professional body warns that ordinary voice banking will not clean up speech that is already slurred or weak, so anything recorded now will carry that. It adds that clinical experience indicates the result may still be acceptable and useful to people, and that some people decide not to use the voice they have made.
“Voice banking with dysarthria or dysphonia will result in a PSV with dysarthria or dysphonia – there will be no 'correction' of any impairment.”
Role of the speech and language therapistLink checked August 2026
Statement 2 of 2. Recordings made before diagnosis can sometimes be used to build a voice, particularly with AI cloning, but only if there is enough good quality material.
Supports this. ALS Voice Banking and Cloning — Your ALS Guide · Practical guide
Gives past recordings such as voice messages left for other people as a genuine alternative to recording new material for AI cloning.
“You can either voice bank for AI cloning ahead of time or use past audio recordings of yourself, such as voice messages you left for others.”
AI Voice CloningLink checked August 2026
Qualifies this. Voice Banking Guide — University of Iowa Speech, Language, and Hearing Clinic · Practical guide
Confirms that services will work from old recordings but treats it as something to be assessed individually, and only where the existing material is good enough.
“You can create a voice based on old recordings on a case-by-case basis if there are sufficient high-quality recordings.”
CereVoice Me, FeaturesLink checked August 2026
Used across the whole answer
#
A speech pathologist is the best person to guide voice banking. A few questions to bring:
Getting started
- Given where my speech is now, is voice banking or message banking worth doing, and how soon?
- Would you suggest voice banking, message banking, voice cloning, or a mix for me?
- What options are available to me, and are any free or supported?
The practical side
- What equipment do I need, and where is best to record?
- How much recording is involved, and can I do it in short sessions?
- Can someone help me, and can I message-bank phrases that matter to my family?
Using it later
- How would a banked voice be used on a communication device?
- Where are my recordings stored, and how do I keep them safe?
- If my voice changes more, will what I have banked still work?
It is fine to start small, to take it at your own pace, and to come back to it. If having your own voice and words preserved matters to you, asking early gives you the most options.
Explained: what this word meansLived experience
Practical tips and experiences shared by people affected by MND. These are not medical advice and may not apply to everyone.
No lived-experience tips for this card yet.
Resources
Your review of the resources
Are these useful, trustworthy, relevant and the right level? Anything missing, too regional, or better on another card?
Notes
Personal Planning Notes
Private to you. Saved on this device only, never sent to us.
Improve this card
A short overall review of this whole card. You can also give feedback on individual answers, resources and steps in their tabs.
Have a resource or a practical tip to share instead? Share what you have learned.