Quick Start

1. Install

python -m pip install rapidly

Rapidly is compatible with Python 3 and later, on Windows and Linux.

2. License

Run instantly with a watermarked output, or contact us for a Rapidly license to unlock the models and remove the watermark.

For details on the licensing model and what's coming next, see the Licensing page.

3. Run your first denoise

import rapidly

engine = rapidly.RapidlyEngine()
engine.add_license("lk_...")  # optional, omit to run in watermarked demo mode

# Fetch the latest Rapidly models
rapidly.update_models()

# Denoise an audio file with the 32 ms model
rapidly.process_file(
    model_file_path="models/speech-denoise-32ms.v1.0.rapidly",
    input_file_path="noisy.wav",
    output_file_path="clean.wav",
)

That's it. clean.wav now contains the denoised audio.

Next steps

• Browse the available models and pick the latency / size that fits your use case.
• Read the full Python API for streaming, per-bus controls, and parameter tuning.
• For native integration, see the C++ interface.
• Confirm your target platform on the Inference page.