AI BriefingResemble AI Unveils DramaBox
·2026.05.15 04:55
Key point
Resemble AI has unveiled DramaBox, a TTS that turns scene descriptions into acted-out speech and even applies watermarking.
1 / 2
Details
DramaBox is a prompt-driven TTS that reads scene descriptions to generate voice with acting tone.
- It operates in a script format where dialogue is written inside double quotes and acting directions are written outside double quotes.
- If a reference voice of 10 seconds or more is provided, it applies that voice's timbre; if not, it generates a new voice matching the description.
- Output is 48 kHz stereo, and generation takes about 2.5 seconds on a warm H100 server.
- All outputs are watermarked by default with Resemble Watermarker (PerTh), and the official page claims about 100% detection accuracy even after MP3/AAC conversion.
- The model is open source and currently English-only. It is available via a Resemble account and on Hugging Face.
- Technically, it is an IC-LoRA fine-tune built on top of the LTX-2.3 audio branch.
Resemble stated that cumulative Hugging Face downloads across its TTS lineup have surpassed 10 million+.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Jevper Released: A Jev Classification Wrapper for OpenAI-Compatible Models with Automatic Logprobs Detection
Rootless Linux Sandbox 'Drop' Released for Isolating AI Agents and Third-Party Programs