mlx-community/parakeet-tdt-0.6b-v3
🧠 AI Modelmlx-community
A high-performance, lightweight speech-to-text model optimized for Apple Silicon using the MLX framework.
The parakeet-tdt-0.6b-v3 model is a specialized port of the Parakeet Transducer-Decoder architecture, specifically tailored for the MLX machine learning framework. At 0.6 billion parameters, it strikes an ideal balance between model size and transcription accuracy, making it highly suitable for on-device deployment. The model utilizes the FastConformer architecture, which combines the strengths of Transformers and Convolutions to capture both local and global context in audio signals efficiently. By using the safetensors format, it ensures secure and fast loading of model weights. This implementation is specifically optimized for Apple Silicon (M-series chips), utilizing unified memory and hardware acceleration to achieve high throughput during real-time speech recognition tasks. It is an essential tool for developers looking to implement privacy-focused, offline voice-to-text features in their native Apple ecosystem applications.
💡Highlights
- ├─0.6B parameter FastConformer
- ├─Optimized for Apple Silicon
- └─High-speed ASR inference
🎯For
- ├─AI Developers
- └─Apple Ecosystem Engineers