Paste a script, pick a voice, get a clean audio file. The PixelDance Audio Studio routes to the top TTS and voice-cloning models — ElevenLabs v3, MiniMax Speech-02 HD, F5-TTS — for narration, character dialog, podcast clips, explainer voiceovers, and short ambient audio.

Output is MP3 or WAV, studio-grade sample rates, downloadable and kept in your library for re-use.

How it works

Pick an engine by use case

ElevenLabs v3 — most expressive, wide voice library, best for narration + character work. MiniMax Speech-02 HD — strong Chinese + English, lower cost. F5-TTS — open-source, cheapest, for bulk production.

Paste or type the script

Up to a few paragraphs per render. Punctuation matters — commas and periods shape pacing. Quote marks trigger emphasis on some models.

Pick voice + tune pacing

Browse the voice library, preview a few samples. Adjust speed and emotion if the engine exposes those controls.

Render + download

Short clips (< 10s) stream instantly. Longer renders queue 15–60s. Download MP3 / WAV. Output stays in your library.

Engines available

ElevenLabs v3 — Best-in-class expressive TTS, largest voice library, strongest on English with good multi-language. Our default for narration / podcast / character work.
MiniMax Speech-02 HD — Strong Chinese + English + more. Lower cost than ElevenLabs. Good for bilingual content creators.
F5-TTS — Open-source, cheapest tier. Voice cloning with a short audio reference. Quality is a notch below flagships but unbeatable for bulk / backend usage.

Prompt tips

Use punctuation deliberately — commas and ellipses create natural pauses.
For emphasis, most engines respect caps or italics hints. ElevenLabs v3 also accepts bracketed emotion tags like [whispering] or [laughing].
Break very long scripts into chapters (< 500 words each) for better pacing continuity.

AI Voice Generator — Questions we hear

Can I clone my own voice?▾

Yes — F5-TTS accepts a short reference audio clip (10–30 seconds) and clones the timbre. ElevenLabs also supports voice cloning with a consented reference. Don't clone voices you don't have permission to use.

What languages are supported?▾

ElevenLabs v3: English, Chinese, Japanese, Korean, Spanish, French, German, Portuguese, and ~20 more. MiniMax: strong on Chinese + English + Japanese. F5-TTS: depends on the reference audio you provide.

What audio formats do I get?▾

MP3 (compressed, universally compatible) and WAV (uncompressed, studio-grade). Both at 44.1kHz or 48kHz sample rates.

Can I use the audio commercially?▾

Yes. You own the output. Underlying model licenses pass through — ElevenLabs commercial use requires appropriate tier, F5-TTS is permissive. Check individual model notes for specifics.

What's the max script length?▾

AI Audio Studio — Voiceover, TTS, and Voice Cloning