Voice Studio
Clone a voice from a recording or design one from a description.
The Voice Studio is where you make voices that aren't in the preset list. There are two ways in — clone a voice from a short audio sample, or design one from a written description — and both end the same way: a reusable voice in your My voices library that you can put on any character.
Opening the Voice Studio
Go to Chat Settings → Voice & Input → Voice Studio and tap Open Voice Studio (the wand icon). Your library is account-wide, so it looks the same from every chat.
Inside you'll find your voice-clone credit balance, the two create actions — Clone a voice and Design a voice — and My voices.
In a group chat the studio is read-only. You can open it and create voices, but the Use action won't assign one. Assign voices to group members from the per-member rows under Chat Settings → Voice & Input → Voice instead.
Cloning from a sample
Open Clone a voice and name it
Names are required and capped at 60 characters.
Pick a provider
MiMo — Free (limited time). MiniMax — 1 credit from your voice-clone balance.
Provide a sample
Record captures one with your microphone; Upload takes an audio file. Aim for 10–30 seconds of clear speech.
Create the clone
Tap Create clone. The voice is ready in the same request — there is no queue to wait through.
What the sample has to be
| Rule | Value |
|---|---|
| Minimum length | 5 seconds, enforced on both recordings and uploads |
| Recommended length | 10–30 seconds |
| In-browser recording cap | 60 seconds — recording stops itself at the limit |
| File size | No limit |
| Format | Anything the browser can read; everything is converted to WAV before upload |
| Speech check | Applied to recordings only |
A recording that is mostly silence is rejected with "We couldn't detect clear speech." An uploaded file skips that check entirely — it is only length-checked — so a quiet or noisy upload is accepted and produces a poor clone. Record clean audio either way: one speaker, no music, no room echo.
If the microphone is blocked, the studio says so and points you at Upload.
Designing from a description
No recording, no sample. Open Design a voice, name it, and write a Voice description of 1 to 1,000 characters — for example, "A warm, gentle young woman with a slight British accent" — then tap Create voice.
Voice design runs on MiMo only and is free for a limited time. MiniMax cannot design voices.
What happens after you submit
Both providers return a finished voice inside the same request, so a new voice appears as Ready straight away. A short preview clip is generated with it — tap play on any Ready voice to hear it. If something goes wrong the voice shows Failed with a short reason, and you can delete it and try again.
| Provider | Clone | Design | Cost to create |
|---|---|---|---|
| MiMo | Yes | Yes | Free during the current promotion |
| MiniMax | Yes | No | 1 voice-clone credit |
Nothing limits how many voices you keep — only your credit balance limits how many MiniMax clones you can make.
Putting a voice on a character
For your own listening, open the voice picker (Chat Settings → Voice & Input → Voice), switch to the My voices tab and select it. That choice is yours alone.
To make it the voice everyone hears, open the character in the editor, go to General Settings → Voice Selection and pick the custom voice. Reverie asks you to confirm first:
Binding a cloned voice publishes how it sounds. Anyone who chats with that character hears the cloned voice by default. They can switch to another voice, and they cannot add yours to their own library — but they can hear and preview it for as long as the character is published.
Voice-clone credits
Voice-clone credits are a separate balance from chat credits: granted plus purchased, minus used. The Voice Studio shows the breakdown at the top.
| Source | Amount |
|---|---|
| Pro, annual | 1 per yearly renewal |
| Premium, annual | 2 per yearly renewal |
| Ultimate, annual | 3 per yearly renewal |
| Monthly plans | None |
| One-time purchase | $5 for one clone |
Only annual plans grant clones. A monthly subscription at any tier grants zero voice-clone credits. Buy the one-time clone instead, or use MiMo while it is free.
Creating a voice is what spends a credit. Playing it afterwards is billed as ordinary text-to-speech at that provider's rate — see Reading messages aloud.
Consent and your samples
Above the create button the studio states: "By continuing you confirm you have the right to use this voice." Clone voices you own or have explicit permission to use.
Your sample is stored in a private bucket and kept until you delete the voice. Deleting a voice removes it from your library and cleans up the stored sample and the copy held by the provider.
Why the studio exists. The thinking behind cloning and designing instead of picking from a list is in The voice was always the hardest part.
Related
- Reading messages aloud — engines, rates and the voice picker
- Voice calls — where a custom voice gets the most use
- Subscriptions — which plans grant clone credits
- Publishing a character — what changes when others can use your character