Fish Audio S2 Pro is a text-to-speech model for natural speech synthesis and voice cloning, suitable for commercial narration, character dialogue, and content that requires a distinct expressive rhythm. Platform guidance emphasizes its expressiveness and sets it as the default model for this entry point. By unifying text, reference voices, and prosody settings, developers can incorporate voiceover production into applications and content workflows.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
Clarify the model, inputs and outputs, and integration method before selecting.
The S2 series can use [bracket] natural-language expression prompts in text, such as [whisper]; actual results should be confirmed by listening. Capability descriptions are based on public model materials and this platform's documentation; actual parameters, outputs, and billing rules are subject to the corresponding API and pricing sections.
Learn about the voice tasks that s2-pro is suited to solve.
Punctuation, pauses, and sentence structure affect vocal performance. S2 Pro is suitable for comparing narration styles in short samples: first confirm emphasis and rhythm, then expand to the full script; the model's expressive positioning does not mean it can precisely execute every emotional marker.
You can either reuse a saved voice through reference_id or use a recording and accurate verbatim transcript for one-time cloning. The former is suitable for recurring characters, while the latter is suitable for one-off voiceover projects; choose one of the two methods.
Use MP3 for web playback or quick review, and use a WAV container for editing. After synchronous completion, retrieve the result through audio_url; longer tasks can also be configured with callbacks to integrate the waiting process into a task list.
Choose based on specific content and delivery method.
Break product selling points, transitions, and endings into short segments, and compare how speaking speed and script wording affect delivery. Produce the full clip after confirming the result, then check duration and pauses against the visuals.
Choose a voice that can be repeatedly used for a fixed character, and check consistency around actual lines. The application handles character arrangement and audio stitching; a single TTS request cannot directly be equated with automatic multi-character production.
Synthesize feature descriptions, brand stories, and guiding copy into a unified voice. Preserve the correspondence between preview scripts and final clips, making it easy to replace sections after copy changes without remaking the entire video.
Compare based on scripts, voice quality, and production cost.
When the model request header is not set, the platform uses s2-pro. For production, it is still recommended to explicitly specify the model to avoid mixing different models between test scripts and subsequent content due to setting changes.
You can compare S2 Pro and S2.1 Pro using the same line and voice, focusing on speech naturalness, generation wait time, and the amount of subsequent revision. When long-form stability matters more, add S1 as a candidate.
Arrange the inputs first, then connect them to the corresponding application workflow.
Prepare a short line, a character voice, and two emotional directions; first select an approved reference_id.
Submit text to /fish/tts, select s2-pro through the model request header, and provide either a reference_id or a single recording reference. After completion, preview with audio_url; for serialized content, save the model, voice, speaking speed, and format.
Design the task directly from the inputs and acceptance priorities below.
Specify model: s2-pro through /fish/tts; you can compare the effects of natural-language expression prompts such as [whisper] in text with ordinary scripts.
Preview whether the delivery sounds natural, whether markers are read aloud, and whether pronunciation is complete; the single voice reference in this entry point does not equal automatic multi-character arrangement.
Understand the synthesis method and delivery scope.
Answers to common questions when using s2-pro.
Yes. This platform's Fish TTS endpoint defaults to s2-pro when the model request header is not set. We recommend explicitly recording the model for production tasks; the latest model officially recommended and this endpoint's default model are two different pieces of information.
Specify s2-pro through the HTTP request header model. It defaults to s2-pro when not specified; the voice reference_id is for voice selection and is configured separately from the synthesis model.
For fixed characters or series content, you can reuse a voice with reference_id; for one-time projects, you can use references to provide a recording and transcript. The two are mutually exclusive, and one-time cloning accepts only one reference sample.
prosody.speed=1.0 indicates the original speed, volume uses dB, and 0 means the volume is unchanged. You can choose mp3, wav, or pcm; the latter two both use WAV containers, and MP3 bitrates can be 64, 128, or 192.
After setting callback_url, first save task_id and started_at, wait for the completion callback, or query by task ID. Only after obtaining audio_url can you proceed to playback or editing.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs