Voice Studio
Your library of voice samples for cloning, and new voices designed from a description when your speech service can.
The Voice Studio (Create › Voice › Voice Studio) holds your voice samples: short recordings that a voice-cloning TTS service speaks with. It also opens from a character's voice settings, with Voice Studio or Design a voice. Not on mobile.
Your voice library
- Add voices: drop audio files, or a zip of voices, anywhere in the studio, or click to choose. You see what each file will become before anything is saved.
- Search N voices finds one by name. Pick one to play it, trim it, or edit its details (Label, Gender, Language) and transcript.
- Trim: set the start and end from the playhead, then Trim, or Reset. Save audio keeps it. Restore original puts back the recording as it was before its first saved trim.
- Transcript: the exact words spoken in the sample, saved beside it as
<voice>.wav.ref_text.txt. Leave it empty to have your Clip Transcription service write it on first use, or generate it now. With no transcription service, write it yourself. - Open in Playground opens the Text To Speech playground with that voice. Only shown when your speech service clones voices.
The folder button opens or copies the folder your voices are kept in.
The studio manages files and works without a cloning service. When the current speech module doesn't clone, it says so: the samples are kept, but nothing speaks with them until you use one that does.
Design a voice
When the speech service can make up a voice, Design a voice appears beside Add voices. The services that can:
- audio.cpp with BreezeTTS 2 (from a description) or OmniVoice (from attributes: gender, age, pitch…) loaded.
- ElevenLabs.
- Voxta Cloud.
- Write a Description (
An old man with a deep, gravelly voice, speaking slowly and warmly), or pick the attributes the model takes. - Edit the Line to say, or keep the default.
- Generate. Each generation is a new voice, played at once and kept in the History.
- Save to the Voice library, or into a character: it goes into the character's assets as its voice sample, with the line as its transcript.
Design a voice from a character's voice settings opens straight here, saving into that character.
The history is gone when you leave the page, so save the voices you want to keep. A saved voice is an ordinary sample: any cloning service can speak with it.