fastrtc_canary

FastRTC Canary

The Real-Time Communication Library for Python with Canary STT

Turn any python function into a real-time audio and video stream over WebRTC or WebSockets.

Installation

Assume you have fastrtc installed - fastrtc

pip install fastrtc_canary

Features

• 🎯 Direct integration with Nvidia's Canary STT model
• 🔄 Seamless compatibility with all FastRTC features
• 🚀 Real-time audio transcription
• 🔌 Drop-in replacement for FastRTC's STT components

Available Models

• nvidia/canary-1b
• nvidia/canary-1b-flash

Supported Languages

• en (English)
• es (Spanish)
• fr (French)
• de (German)

Quick Start

from multiprocessing import freeze_support
from fastrtc_canary import get_stt_model, load_audio

def main():
    # Load model
    model = get_stt_model(
        model="nvidia/canary-1b",
        lang="en",
    )

    # Load audio file (automatically resamples to 16kHz)
    audio = load_audio("test_file.wav")

    # Transcribe
    text = model.stt(audio)
    print(f"Transcription: {text}")

if __name__ == '__main__':
    freeze_support()
    main()

Note: This is additional package for fastrtc to support Nvidia's Canary STT models. If you want to use fastrtc, you can install it from here

Documentation

For more detailed information about the base FastRTC package, visit https://fastrtc.org

License

MIT