How to voice a text with different voices: dialogue and reading by parts
To voice a text with several voices, mark it up by part, describe each character’s voice in words and ask the neural network to read the dialogue. In the Twelver chat two voices come out in one file from one request, and scenes with three or more characters are assembled from parts into one MP3.
Can a text be voiced with several voices?
Yes. A speech-synthesis model can read a dialogue in two different voices in one recording: you give it the text with speaker names, and it switches voice on every line and keeps the pauses between them. For a third character and beyond, the scene is split into parts, each is voiced separately and the parts are joined into one file. Voices are not picked from a catalogue — you describe them.
A two-voice dialogue in 4 steps
- Mark the text up by part. Each line on its own row, starting with the speaker’s name and a colon. Narration is its own part, “Narrator”.
- Describe each voice in one phrase — age, timbre, manner: “the fox — female, silky, slightly mocking”.
- Set the delivery where it matters. Put a whisper, a shout or excitement into the voice description in words: “the fox whispers, conspiratorially”. Laughs, sighs and pauses go in as tags inside the line — they come out as sound, not words:
<laugh>,<sigh>,<short pause>. - Send it all in one message. Listen, and if a voice is off, fix the description in words: “make the fox sound older”.
How to describe a character’s voice
| Character | How to describe the voice |
|---|---|
| Storyteller, narrator | “Male, mature, low and warm, unhurried, with a smile” |
| Villain | “Male, low, slow, coldly polite, with pauses” |
| Small animal, fairy creature | “High, bright, fast, delighted” |
| Robot, AI | “Even, emotionless, crisp diction, short pauses” |
| Podcast host | “Female, 30, energetic, conversational, no announcer pomp” |
A cartoon or game character’s voice
The model will not copy a real actor’s, blogger’s or celebrity’s voice — it is someone else’s. It will make an original voice in the same spirit: describe how the character sounds and you get a similar character without a forgery. For a child character, or when a description does not land, act the character yourself — twenty seconds in character clone into a voice that sounds exactly as you meant; see a character voice for narration.
Three or more voices: assembling a scene
| What you are voicing | Requests | How it is assembled |
|---|---|---|
| A two-person dialogue, interview, sketch | 1 | Done at once, one file |
| A tale: narrator + two characters | 2–3 | Characters as a pair, narrator separately, stitched |
| A radio play, 4–6 parts | one per scene | Scenes in pairs, stitched in order |
| A text over 7,000 characters | per 7,000 characters | Parts in order, stitched into one MP3 |
Frequently asked questions
How do I voice a text with different voices?
How many voices can one voice-over have?
Can I try it for free?
Can I use the voice-over commercially?
How single-voice narration works and what makes it sound alive is in Voice over text.
Try it yourself
Everything in this guide runs inside Twelver
One chat for text, images, video, music and voice — no separate services or subscriptions.
Open Twelver chat