# babelscribe

Runs offline speech recognition and builds subtitles locally on your GPU. Lets coding agents transcribe local audio and video files across 99 languages without cloud APIs.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 6
- Author: phonology024
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/phonology024/babelscribe
- Updated: 2026-10-10
- HTML: https://affpapa.org/en/nevibecode/mcp/babelscribe

## What it does

The server wraps `whisper.cpp` to deliver transcription and subtitle generation directly on local hardware. It runs on AMD, NVIDIA, Intel, and Apple Silicon chips using Vulkan, CUDA, or Metal, with a CPU fallback when no GPU is present. Because it fetches precompiled `whisper-cli` binaries automatically, you avoid building native drivers manually.

Four tools are exposed to the AI model: `transcribe` for making subtitle and text files, `find_media` to locate recent media in user directories, `list_devices`, and `list_languages_and_models`. In Codex, default timeout limits cause jobs to drop during the initial model fetch. You must raise `tool_timeout_sec` to 3600 in your Codex configuration file before running large files.

## Requirements
- Needs: Python and uv (commands run via uvx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add babelscribe -- uvx babelscribe
```
uvx runs the command named after the package; if the README names another one, use uvx --from babelscribe <command>

### Codex

**Platform CLI**
```bash
codex mcp add babelscribe -- uvx babelscribe
```
uvx runs the command named after the package; if the README names another one, use uvx --from babelscribe <command>

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "babelscribe": {
            "command": "uvx",
            "args": [
                "babelscribe"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from babelscribe <command>

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "babelscribe": {
            "type": "stdio",
            "command": "uvx",
            "args": [
                "babelscribe"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from babelscribe <command>

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add babelscribe uvx babelscribe
```
uvx runs the command named after the package; if the README names another one, use uvx --from babelscribe <command>

## When to use it
- generate SRT or VTT subtitles from video recordings
- transcribe speech locally without uploading media to third-party clouds
- locate the newest video in standard system folders automatically
- run Whisper models on AMD graphics cards using Vulkan

## Example prompts
- Find the newest recording in my Downloads folder and generate subtitles for it.
- Transcribe interview.mp4 and write the text transcript next to the file.
- List all available hardware devices for speech processing.

## Similar
- [MCP Toolbox for Databases](https://affpapa.org/en/nevibecode/mcp/googleapis-mcp-toolbox-for-databases.md) — MCP server · ★ 16730
- [labelhead-artist-momentum](https://affpapa.org/en/nevibecode/mcp/labelhead-artist-momentum.md) — MCP server · ★ 107701
- [Agent Skills Search Server](https://affpapa.org/en/nevibecode/mcp/agent-skills-search-server.md) — MCP server · ★ 26673
- [Skyvern](https://affpapa.org/en/nevibecode/mcp/skyvern.md) — MCP server · ★ 23283
- [Nhost](https://affpapa.org/en/nevibecode/mcp/nhost.md) — MCP server · ★ 9369
- [PraisonAI](https://affpapa.org/en/nevibecode/mcp/praisonai.md) — MCP server · ★ 9298

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md