TranscribeTools 0.5.4

Transcribe all soundfiles in a folder using Whisper

LocalWhisper

Introduction

TranscribeTools contains an Python application LocalWhisper which transcribes all sound files in a configurable folder using a local Whisper model. You can choose which Whisper model is to be used

Details

• using Python 3.12.7, openai-whisper https://pypi.org/project/openai-whisper/ (current version 20240930) does not support 3.13 yet.

License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Setup

We use uv for managing virtual environments and package installation. Follow these steps to set up the project:

On macOS:

Install uv

• First install brew if needed from https://github.com/Homebrew/brew/releases/latese

On Windows:

Download the setup script

We need to install UV a tool to install the Python environment and to install the tool. There are a few possibilities

1. Follow instructions at the UV website

2. Press {Windows button} then type or paste:. powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

3. Use winget

• Open the Windows Powershell: press {Windows} button and type or paste

winget install --id=astral-sh.uv  -e

These scripts will:

Installs the uv tool. Check if uv {enter} works. At the moment a reboot is needed on Windows. Now we can install the tools.

Install tools

uv tool install transcribetools

Install the (commandline) tools in this project. For now it's only transcribe_folder.

Plans

• Make it a local service, running in the background
• Investigate options to let it run on a central computer, as a service
• Create Docker image
• Add speaker partitioning (see TranscribeWhisperX)
• Adjust models using PyTorch (more control)

Documentation about Whisper on the cloud and local

• Courtesy of and Credits to OpenAI: Whisper.ai
• doc
• alternatief model transcribe and translate