Fish Audio TTS
Fish Audio S2.1‑Pro speech with adjustable reading speed
Bring a Fish Audio voice model ID — designed or cloned — and S2.1-Pro reads your script at 0.5–2.0× speed with the language detected from the text. Credits bill per UTF-8 byte of input, quoted before you generate.

Model Overview
Fish Audio S2.1-Pro
Fish Audio TTS runs the production S2.1-Pro model on Moosky: hand it a script and a Fish Audio voice model ID — one saved from Voice Design or from a completed voice-clone enrollment — and it returns finished MP3 narration.
The details
Prosody is tunable per run: the speed slider goes from half to double the default rate, and the model detects the script's language automatically, so mixed or foreign-language copy reads without a manual switch. The optional language hint only nudges ambiguous cases.
Billing follows the text, not the audio: S2.1-Pro is priced per UTF-8 byte of your script and the credit cost is quoted before you generate, so you can price a long audiobook chapter or a two-line ad without surprises.
Why use it on Moosky AI?
Model-specific controls with no subscription, clean outputs, and a workflow built for fast creative iteration.
Production S2.1-Pro voice
The flagship Fish Audio synthesis model, available on Moosky for per-run credit billing — without a separate Fish subscription.
Tune the reading speed
Slide prosody from 0.5× to 2.0× per run — stretch a tutorial for clarity or double-time a hype read without re-recording.
Bring your own voice ID
Paste a Fish Audio voice model ID saved from Voice Design or a clone enrollment; every run reuses that exact timbre.
Language detected from the script
S2.1-Pro reads the language out of your text automatically; the optional hint only resolves ambiguous cases.
How it works
Three focused steps from setup to finished output.
1. Bring a voice model ID
Paste the ID of a Fish Audio voice you designed or cloned — it drives every run from here.
2. Set speed and script
Type the narration and nudge the 0.5–2.0× prosody slider; language detects itself from the text.
3. Quote, generate, download
The per-byte credit cost shows first; the finished MP3 lands in your library for any video or bed.
Credit-based
Simple, honest pricing
Billed per UTF-8 byte of input text for the S2.1-Pro model, with the credit cost quoted before you generate. No subscription.
More from Moosky AI
Explore our full suite of AI tools.
Frequently Asked Questions
Where do I get a Fish Audio voice model ID?
Two routes on Moosky: design a voice with Fish Audio Voice Design, or enroll a clone from a reference clip on the Fish Audio Voice Clone page. Both return a model ID — paste it into the Voice Design ID field and S2.1-Pro synthesizes in that voice.
Is Fish Audio TTS billed by audio length?
No — the S2.1-Pro model bills per UTF-8 byte of the input script, not the generated file. A ten-line ad and a long-form chapter are priced by how much you typed, and Moosky shows the exact credit figure before the run starts.
Does the speed slider change pitch or just pace?
It changes speaking rate, from half the default pace to double. The voice's identity stays put, so you can speed up an explainer or slow down a language lesson without the timbre wobbling.
Do I have to set the language before generating?
No. S2.1-Pro auto-detects the language of your script — matching the broad multilingual coverage Fish Audio publishes for the model — and the optional hint field exists only for edge cases where the text is ambiguous.
What audio format comes back?
Each run returns an MP3 rendered by the S2.1-Pro engine — ready to drop into editors, video beds, or podcast drafts without a conversion step.
Can I reuse one voice ID across projects?
Yes — the voice model ID is a saved profile, not a session token. Point it at an explainer today and an audiobook next month and both render in the same timbre, billed byte-for-byte per run.
