drama-3: Generate multi-speaker audio (35B, 3B active)
Added · Updated
Made by Fish Audio
View profile on Hugging Face (opens in a new tab)Turn text into multi-speaker speech with voice cloning and continuation controls.
For audio creators using one of the documented inference servers.
Model details
Original model fishaudio/drama-3
Input Text Output Audio
Ways to use it Run locally
What it does
Fish Audio Drama 3 turns text into speech for multi-character scenes and long-form content. The model card reports 35B total and 3B active parameters per token, and lists voice cloning, speaker tags, natural-language controls, and audio editing. Long text is chunked automatically and can be streamed per chunk. The examples also show continuation from earlier audio tokens so a changed line can be generated with previous takes in context. The model is served through SGLang-Omni or vLLM-Omni; it is not a ready-made browser audio editor.
How to use it
Run the documented SGLang-Omni or vLLM-Omni server, then send script text and supported speaker or reference-audio inputs to its text-to-speech endpoint to receive generated audio.
Common use cases
- Send dialogue with speaker tags and receive a multi-character audio scene.
- Provide reference audio and its transcript, then submit narration text for a voice-cloning workflow.
- Send corrected dialogue with previous takes and receive an edited continuation.
Can I use this?
Requirements The card documents SGLang-Omni or vLLM-Omni setup, but does not give a minimum GPU-memory requirement. The Fish Audio Community License permits research and non-commercial use and limited commercial use subject to its revenue threshold; hosted speech services require a separate written agreement. It also requires consent to imitate an identifiable person’s voice.
Caveat The Fish Audio Community License restricts hosted services and voice imitation without consent. Commercial use is limited by a stated revenue threshold; read the license for all conditions.
Worth knowing Fish Audio Community License permits research and non-commercial use and limited commercial use below a stated annual-revenue threshold; a hosted speech service needs a separate written license. Voice imitation requires consent, and use restrictions apply.
Why it's here
Not trending this week
Last seen on the Hugging Face trending list on 28 October 2026. It stays on BuildTube so you can still find it.
- 1,226 people have liked it on Hugging Face.
- It was downloaded 43,735 times in the last 30 days.
Numbers checked on 8 November 2026; not refreshed since.
Behind it
View model pageBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is drama-3?
- Fish Audio Drama 3 turns text into speech for multi-character scenes and long-form content.
- What can I use drama-3 for?
- Send dialogue with speaker tags and receive a multi-character audio scene; Provide reference audio and its transcript, then submit narration text for a voice-cloning workflow; Send corrected dialogue with previous takes and receive an edited continuation
- Can I run drama-3 on my computer?
- It runs on your own hardware with a model runtime. The card documents SGLang-Omni or vLLM-Omni setup, but does not give a minimum GPU-memory requirement.
- What do I need to use drama-3?
- SGLang-Omni or vLLM-Omni. Its license restricts hosted speech services and certain commercial uses.
- What license does drama-3 use?
- Fish Audio Community License permits research and non-commercial use and limited commercial use below a stated annual-revenue threshold; a hosted speech service needs a separate written license. The Fish Audio Community License restricts hosted services and voice imitation without consent.
More from this week
- P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
- Idol Turn one picture of an object into a 3D Gaussian splat
Idol turns one photo, render or generated image of an object into a 3D Gaussian splat file.
- agent-desktop Let AI agents read and click desktop apps
For developers building computer-use agents: agent-desktop lets an AI agent read and operate desktop apps on Mac, Windows or Linux from a command line.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/hf-fishaudio-drama-3/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/hf-fishaudio-drama-3/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="drama-3 on BuildTube"></iframe>