Fish Audio announced the launch in a press release. The company’s growth figures are its own claims, and the release gives no independent benchmarks.
Fish Audio, a voice AI company based in Palo Alto, California, has launched Drama 3, a text-to-speech model. Text-to-speech means software that turns written text into spoken audio [1].
The company says Drama 3 lets people “direct” a voice instead of just generating one. Users describe the tone, pacing and character they want in plain language, then adjust the performance line by line [1].
What Drama 3 does
Fish Audio lists four features [1]:
- Plain-language direction. You describe a performance in words, and can change it mid-sentence, so a line can start calm and turn panicked without recording a new take.
- Several speakers at once. A whole scene can be generated in one go, with each character keeping its own voice and reacting to what the other just said.
- Audio inpainting. You pick a part of the audio to change, update the instructions, and Drama 3 “regenerates only that segment, keeping everything around it intact.” This is like fixing one word without re-recording the whole clip.
- Long scripts and many languages. The voice stays consistent across long scripts, can switch language mid-scene, and lets users control pronunciation sound by sound (at the “phoneme” level) “in English, Chinese, Japanese, Korean, and more.”
Fish Audio says it is aimed at audio dramas, audiobooks, dubbing, games, animation and AI companion apps [1].
“Voice AI has gotten very good at generating speech. The next step is directing a performance,” said Rissa Cao, co-founder and CEO of Fish Audio [1].
“Drama 3 adopts a new architecture that allows for better control and instruction following,” said Shijia Liao, co-founder and chief scientist [1]. The release does not describe that architecture.
Price and availability
Drama 3 is available now at fish.audio. For the first 30 days, use through its API (the interface developers connect to) is 80% off, and businesses switching from another voice provider get their first three months free [1].
The company’s claims
Fish Audio says [1]:
- it has grown “from zero to $21 million in annual recurring revenue” and serves “more than 15 million users”;
- it ranked No. 1 on Ahrefs’ list of the 50 fastest-growing AI companies, “with brand search up 740% year over year”;
- Drama 3’s predecessor, S2.1 Pro, “became the most-requested speech model on OpenRouter within three weeks of launch.”
TSN has not independently checked these figures.
What this means
- Voice AI is moving towards editing, not just generating. Drama 3’s pitch is control: fixing single words and changing emotion mid-line.
- No outside tests yet. The release gives no independent benchmarks, so how well Drama 3 performs against rivals is not yet known.
- The growth figures are Fish Audio’s own.
Sources
- Fish Audio via PR Newswire, “Fish Audio Launches Drama 3: Direct AI Voices in Plain Language, Control Every Word and Shift Emotion Mid-Sentence”, 13 October 2026. https://www.prnewswire.com/news-releases/fish-audio-launches-drama-3-direct-ai-voices-in-plain-language-control-every-word-and-shift-emotion-mid-sentence-302906620.html
