Skip to content

Releases: avivshaked/narration-mcp

Release list

0.1.0a1: the first alpha, for testers

0.1.0a1: the first alpha, for testers Pre-release
Pre-release

Choose a tag to compare

View all tags
@avivshaked avivshaked released this 30 Sep 10:37
v0.1.0a1
7d7564d

The first tagged release of narration-mcp: an alpha, for testers.

narration-mcp is a local MCP server that gives an AI agent narration in one fixed voice. It designs a voice with Qwen3-TTS, measures it once, and narrates paragraphs of text by cloning that voice. Every take comes back checked (what was said, who said it, how fast) and aligned (the time of every line within the audio).

It has run on one machine so far. If you have a Windows 10 or 11 PC with an NVIDIA GPU (about 12 GB of VRAM free) and about 25 GB of disk, please install it and tell us where it breaks: see Help test it, and report with a tester report.

git clone --branch v0.1.0a1 https://github.com/avivshaked/narration-mcp.git

Known limits of this alpha

  • Windows only, and an NVIDIA GPU is required.
  • The accuracy of the cue times has not been measured yet.
  • audition_pronunciation has been tested only against the service's fake workers, not on a GPU.
  • Schemas and configuration keys may change before 1.0; CHANGELOG.md lists every change.

Licence: free for personal and other non-commercial use (PolyForm Noncommercial 1.0.0); commercial use, including testing inside a company, needs a licence (COMMERCIAL.md). The models have licences of their own.

Loading