jun7799/scribe-transcribe

Analysis updated 2026-06-24

★ 40PythonAudience · generalComplexity · 3/5LicenseSetup · moderate

Mindmap

mindmap
  root((scribe-transcribe))
    Inputs
      WeChat Channels URL
      Local browser session
      ffmpeg audio
    Outputs
      MP4 download
      SRT subtitles
      Plain text transcript
    Use Cases
      Archive WeChat videos
      Make subtitles offline
      Bulk transcription
    Tech Stack
      Python
      Go
      faster-whisper
      ffmpeg

mindmap root((scribe-transcribe)) Inputs WeChat Channels URL Local browser session ffmpeg audio Outputs MP4 download SRT subtitles Plain text transcript Use Cases Archive WeChat videos Make subtitles offline Bulk transcription Tech Stack Python Go faster-whisper ffmpeg

Click or tap to explore — scroll the page freely

What do people build with it?

USE CASE 1

Download a WeChat Channels video from the desktop browser by clicking the injected button.

USE CASE 2

Transcribe an existing MP4 to SRT and plain text using local faster-whisper without an API key.

USE CASE 3

Run a phone-side WeChat in-app download by pointing the device proxy at the desktop machine.

USE CASE 4

Switch faster-whisper from CPU to CUDA in config.py to speed up long video transcription.

What is it built with?

PythonGofaster-whisperffmpeg

How does it compare?

	jun7799/scribe-transcribe	rbrown101010/codex-marketing-skills	tokyubevoxelverse/forzadesigner6
Stars	40	40	40
Language	Python	Python	Python
Setup difficulty	moderate	moderate	moderate
Complexity	3/5	2/5	3/5
Audience	general	writer	general

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Needs Go 1.21+, Python 3.10+, ffmpeg, and a 462 MB faster-whisper model download on first run.

GPL-3.0-or-later means you can use and modify the code but any redistributed version, including network-served forks, must stay open source under the same license.

In plain English

This is a Chinese-language command-line tool that does two things together: it downloads videos from WeChat Channels, the Tencent short-video service, and then transcribes the audio into subtitles and plain text. The README is in Chinese and the workflow assumes a user who already opens WeChat Channels in a desktop browser. The tool is a Python wrapper around a Go program. The Go binary, called wx-dl, runs a local proxy and an HTTP API. When the user opens the WeChat Channels web page through this proxy, the page receives an injected download button that the original site does not show. The Python side, main.py, manages the proxy process, calls ffmpeg to pull a 16 kHz mono WAV out of the downloaded file, runs the audio through faster-whisper for local speech-to-text, and exports SRT subtitles plus a plain text version. To install, the user needs Go 1.21 or newer, Python 3.10 or newer, ffmpeg, and git. They clone the repository, build the Go downloader into a bin folder, then run pip install -r requirements.txt. A Tsinghua mirror URL is offered for users in mainland China. The first transcription downloads a faster-whisper model of roughly 462 MB and caches it for later runs. Day-to-day usage is start.bat on Windows or start.sh on Mac and Linux. After the script launches, the user opens channels.weixin.qq.com in a normal desktop browser, and the injected button lets them save the current video. There is no need to configure a proxy on the phone for this path, the proxy lives on the same machine that runs the browser. The README also describes a CLI mode with python main.py serve, python main.py transcribe some-video.mp4, and python main.py tasks. An optional section explains how to point a phone at the desktop's proxy through Wi-Fi for in-app downloads. The FAQ covers a missing download button, mobile proxy connectivity, slow CPU transcription that can switch to CUDA by editing WHISPER_DEVICE in config.py, and Windows or sudo permissions. The license is GPL-3.0-or-later, matching the upstream scribe-studio project. Credits list scribe-studio and wx_channels_download.

Copy-paste prompts

Prompt 1

Install scribe-transcribe on Windows: install Go 1.21, build wx-dl, pip install -r requirements.txt, then run start.bat.

Prompt 2

Use scribe-transcribe CLI mode to transcribe an existing MP4 file with python main.py transcribe some-video.mp4.

Prompt 3

Configure scribe-transcribe to use my CUDA GPU by editing WHISPER_DEVICE in config.py and verify the speedup.

Prompt 4

Set up a mobile phone over Wi-Fi to download WeChat Channels videos through the scribe-transcribe local proxy.

Frequently asked questions

What is scribe-transcribe?

Chinese CLI that downloads WeChat Channels videos through a local proxy then transcribes the audio to SRT and plain text with faster-whisper.

What language is scribe-transcribe written in?

Mainly Python. The stack also includes Python, Go, faster-whisper.

What license does scribe-transcribe use?

GPL-3.0-or-later means you can use and modify the code but any redistributed version, including network-served forks, must stay open source under the same license.

How hard is scribe-transcribe to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is scribe-transcribe for?

Mainly general.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.