FMCV Transcripts

Universal Broadcast Transcripts System

If you find this project useful, consider buying me a drink!

---

Quick Start

1. Clone the repository:

   git clone https://github.com/cyysky/FMCV_Transcripts

2. Copy the .env file:

   cp .env.example .env

3. Download and install Docker Desktop:
   Docker Desktop

4. Install ollama
   Ollama
   download ollama model
   ollama run deepseek-r1:8b-llama-distill-q8_0
   ollama run deepseek-r1:1.5b

---

Configuration

For NVIDIA GPU (16GB VRAM or higher)

Recommended .env settings:

MIN_PROCESS_GAP_SECONDS = 0.5  
MAX_PROCESS_TIME_MINUTES = 1  

Start the application with GPU support:

docker compose --profile gpu up -d

For CPU (Experimental)

Recommended .env settings:

MIN_PROCESS_GAP_SECONDS = 30  
MAX_PROCESS_TIME_MINUTES = 100  

WHISPER_MODEL=openai/openai/whisper-tiny
OPENAI_MODEL=deepseek-r1:1.5b

Start the application with CPU support:

docker compose --profile cpu up -d

---

Usage

Access the application at:
http://localhost:8999

---

Introduction Video

Watch the introduction video for more details:

Introduction Video

Enjoy using FMCV Transcripts! 🚀

Contact : [email protected]⁠