fish-drama-3Fish Audio Drama 3 is a text-to-speech model built for acted speech rather than flat narration, released by Fish Audio on October 13, 2026. You direct it from inside the text: a bracketed note such as [barely holding it together, forcing a smile] before a line changes how that line is delivered, and paired tags such as <whisper>, <emphasis>, <soft>, <fast>, <slow>, <stress> and <forceful> change only the words they wrap. Directions can be written in English or in your own language. One request can voice a whole scene: pass reference_id as an array of voice ids and switch speakers with <|speaker:0|>, <|speaker:1|> and so on, up to 16 speakers. Names and terms can be pinned with phoneme tags, using CMU Arpabet for English, pinyin for Chinese and romaji for Japanese. It covers 83 languages, and Fish Audio cites 93.2% pronunciation accuracy on Artificial Analysis against 88.1% for S2.1 Pro. Billing is per 1,000 UTF-8 bytes, the same unit as S2.1 Pro, so a Chinese character counts as three bytes. The price is $0.065 per 1,000 bytes, with a special price of $0.025 until November 12, 2026. In our test, 1,000 English characters took 9.6 seconds and returned 89 seconds of audio. For plain narration that needs no direction, S2.1 Pro at $0.03 per 1,000 bytes is the better fit. Fish Audio's licensed voices (marked licensed in the voice library) are billed at 2.5 times the rate: $0.1625 per 1,000 bytes, or $0.0625 during the special. Community voices and your own clones stay at the normal rate.
[forcing a smile] before a line, or <whisper>…</whisper> around words; directions can be written in any language
Pass reference_id as an array and switch with <|speaker:N|>; speaker 0 uses the first voice in the array
Phoneme tags for names and terms: CMU Arpabet for English, pinyin for Chinese, romaji for Japanese
Special price $0.025 until Nov 12. Billed per byte, so a Chinese character counts as three
Fish Audio Drama 3 is an Audio & Speech API provided by Fish Audio. Fish Audio Drama 3 is a text-to-speech model built for acted speech rather than flat narration, released by Fish Audio on October 13, 2026. You direct it from inside the text: a bracketed note such as [barely holding it together, forcing a smile] before a line changes how that line is delivered, and paired tags such as <whisper>, <emphasis>, <soft>, <fast>, <slow>, <stress> and <forceful> change only the words they wrap. Directions can be written in English or in your own language. One request can voice a whole scene: pass reference_id as an array of voice ids and switch speakers with <|speaker:0|>, <|speaker:1|> and so on, up to 16 speakers. Names and terms can be pinned with phoneme tags, using CMU Arpabet for English, pinyin for Chinese and romaji for Japanese. It covers 83 languages, and Fish Audio cites 93.2% pronunciation accuracy on Artificial Analysis against 88.1% for S2.1 Pro. Billing is per 1,000 UTF-8 bytes, the same unit as S2.1 Pro, so a Chinese character counts as three bytes. The price is $0.065 per 1,000 bytes, with a special price of $0.025 until November 12, 2026. In our test, 1,000 English characters took 9.6 seconds and returned 89 seconds of audio. For plain narration that needs no direction, S2.1 Pro at $0.03 per 1,000 bytes is the better fit. Fish Audio's licensed voices (marked licensed in the voice library) are billed at 2.5 times the rate: $0.1625 per 1,000 bytes, or $0.0625 during the special. Community voices and your own clones stay at the normal rate. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: per 1K UTF-8 bytes: $0.065, Fish licensed voice, per 1K UTF-8 bytes: $0.1625.
Generate professional-grade voiceovers for videos, animations, and ads with diverse voice options.
Quickly produce podcast audio content with support for multi-character dialogue.
Convert text content into natural, fluid speech for audiobook production.
AI-powered multilingual dubbing and translation to help content reach global audiences.
Fish Audio Drama 3 is available through APIMODELS at: per 1K UTF-8 bytes: $0.065, Fish licensed voice, per 1K UTF-8 bytes: $0.1625. Billing is pay-as-you-go — you only pay for what you generate.
Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.
APIMODELS offers the same Fish Audio Drama 3 model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
Write the direction into the text. Put a bracketed note before the line, such as "[barely holding it together, forcing a smile] I'm fine.", and the whole line is delivered that way. To change only a few words, wrap them in a paired tag — <whisper>, <emphasis>, <soft>, <fast>, <slow>, <stress> or <forceful> — for example "Really. <whisper>Go on without me.</whisper>". Directions can be written in English or in the language of the script. Fish Audio also offers language-switch tags such as <french>…</french>, but it labels them experimental: they are not guaranteed for every voice, and the model may switch language on its own.
Pass reference_id as an array of voice ids and mark each turn with <|speaker:N|>, where N is the position in that array: <|speaker:0|> uses the first voice, <|speaker:1|> the second, up to 16 speakers. The ids can come from the Fish voice library or from your own clones. A single string still works when there is only one voice. Arrays are accepted on fish-drama-3 only; fish-s2.1-pro returns a 400 if you send one.
Drama 3 costs $0.065 per 1,000 UTF-8 bytes, with a special price of $0.025 until November 12, 2026. It is billed per byte, like S2.1 Pro: a Latin letter is one byte and a Chinese character three, so 1,000 English characters cost $0.065 and 1,000 Chinese characters (3,000 bytes) cost $0.195 — $0.025 and $0.075 during the special. S2.1 Pro costs $0.03, so for plain narration that needs no direction it is the better fit. Texts under 100 bytes are charged as 100 bytes, and failed requests are not charged.
On APIMODELS, Fish Audio Drama 3 runs alongside 60+ models on one API key and one balance, so choosing is about fit, not lock-in. It supports Inline Directions, Up to 16 Speakers, Pronunciation Tags, 83 Languages, and you can weigh it on price and capability against other Audio & Speech models, then switch by changing a single model-name string — no new account or integration. Browse every Audio & Speech option with live pricing at apimodels.app/models.
Fish Audio Drama 3 supports: Inline Directions, Up to 16 Speakers, Pronunciation Tags, 83 Languages. See the APIMODELS docs for full parameters and call examples.
Yes. APIMODELS exposes Fish Audio Drama 3 through a single unified API and one key — no separate provider accounts, and no need to handle each provider's regional network access yourself.
We support Stripe (Visa, Mastercard, and other international cards) and Alipay. Credits are available instantly after payment.