Metadata-Version: 2.4
Name: transcribe_everything
Version: 1.5.36
Summary: transcribe transcriptions ai whisper
Home-page: https://github.com/zackees/transcribe-everything
Maintainer: Zachary Vorhies
License: BSD 3-Clause License
Keywords: ai,transcribe,anything,everything,s3,remote,big,data
Classifier: Programming Language :: Python :: 3
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: langdetect>=1.0.9
Requires-Dist: transcribe-anything>=3.0.7
Requires-Dist: virtual-fs>=1.0.23
Dynamic: home-page
Dynamic: license-file
Dynamic: maintainer

# transcribe-everything

[![Build Docker Image](https://github.com/zackees/transcribe-everything/actions/workflows/build_docker_image.yml/badge.svg)](https://github.com/zackees/transcribe-everything/actions/workflows/build_docker_image.yml)

Transcribes everything! Point this solution to a remote directory and this tool will find all the media files (*.mp3, *.mp4) and if there is no *.txt present, it will be transcribed. Will continue until all files are transcribed.

# Docker
  * Install
    * docker pull niteris/transcribe-everything

  * Help
    * docker run --rm -it niteris/transcribe-everything --help

--gpu-batch-size

  * Running (easy)
    * Install `transcribe-everything` locally and use the `transcribe-everything-run-docker` tool.
    * `uv venv`
    * `uv pip install transcribe-everything`
    * `uv run transcribe-everything-run-docker --gpu-batch-size 8 --gpu-jobs 1`
      * Play `--gpu-batch-size` and `--gpu-jobs` for performance tuning.
      * Defaults are tested to run stable for Nvidia 3070 12gb card.

  * Running (Manually)
    * Windows cmd.exe: `docker run --rm -it -v "%cd%\rclone.conf:/app/rclone.conf" --gpus all niteris/transcribe-everything dst:TorrentBooks/podcast/dialogueworks01/youtube`
    * Macos/Linux: `docker run --rm -it -v "$(pwd)/rclone.conf:/app/rclone.conf" --gpus all niteris/transcribe-everything dst:TorrentBooks/podcast/dialogueworks01/youtube`
