Japanese Meeting Summarizer

A tool for automatically transcribing and summarizing Japanese meetings (audio/video) using LLM and speech models.

Models Used

1. Speaker Diarization

• pyannote/speaker-diarization-3.1

2. Speech-to-Text (Japanese ASR)

• kotoba-tech/kotoba-whisper-v2.2

3. LLM (Summarization)

• Qwen/Qwen2.5-7B

Installation Requirements

• Linux ONLY
• Node.js (latest LTS recommended)
• Python environment with uv

wget -qO- https://astral.sh/uv/install.sh | sh

• ffmpeg installed

sudo apt install -y ffmpeg

Setup Project

Install step by step:

git clone https://github.com/yusteafy/meeting-summarizer
cd meeting-summarizer

cd frontend
npm install

cd ../backend
uv sync

Run Project

Option 1: Manual run

Start backend:

cd backend
uv run main.py

Start frontend (in a new terminal):

cd frontend
npm run dev

Option 2: One-click script

bash start.sh