lecsum

Automatically transcribe and summarize lecture recordings completely on-device using AI.

Environment Setup

Install Ollama.

Create a virtual Python environment:

python3 -m venv .venv

Activate the virtual environment:

source .venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Configuration (optional)

Edit lecsum.yaml:

Field	Default Value	Possible Values	Description
whisper_model	"base.en"	Whisper model name	Specifies which Whisper model to use for transcription
ollama_model	"llama3.1:8b"	Ollama model name	Specifies which Ollama model to use for summarization
prompt	"Summarize: "	Any string	Instructs the large language model during the summarization step

Usage

Run the Ollama server:

ollama serve

Command-line

In a new terminal, run:

./lecsum.py -c [CONFIG_FILE] [AUDIO_FILE]

Use any file format supported by Whisper (mp3, mp4, wav, webm, etc.).

Server

To start the lecsum server in a development environment, run:

fastapi dev server.py

References

• https://pyyaml.org/wiki/PyYAMLDocumentation
• https://github.com/openai/whisper
• https://github.com/ollama/ollama-python
• https://fastapi.tiangolo.com