#Parakeet ASR Demo

Speech recognition using NVIDIA's Parakeet TDT model.

#Features

• 🎙️ Speech-to-text transcription using NVIDIA Parakeet TDT model
• 📊 Real-time transcription with progress tracking
• 📝 Support for multiple audio formats (WAV, FLAC)
• 📈 Transcription history with export options
• 🎯 Optimized for both short and long audio files
• 💻 GPU acceleration support with fallback to CPU

#Setup

1. Install dependencies: bash pip install -r requirements.txt

2. Run the application: bash streamlit run app.py

#Requirements

• Python 3.8+
• NVIDIA GPU with CUDA support (strongly recommended for optimal performance)
• FFmpeg (for audio processing)

#Usage

1. Upload an audio file or record directly in the browser
2. Wait for the model to process and transcribe
3. View and export transcription results

#Notes

• NVIDIA GPU with CUDA support is strongly recommended for optimal performance
• Long audio files (>8 minutes) will automatically use optimized settings
• Maximum recommended audio duration is 30 minutes