Guides

Capture-side processing

Cleaning audio you produce locally, before it's encoded, sent, or recorded.

Capture-side processing runs Rapidly on the audio coming into your app: a microphone, a line input, a USB capture device. The engine sits between the input and whatever consumes it (encoder, recorder, network, file). Noise, reverb, and unwanted sounds are removed before the audio leaves your hands.

Where the engine sits in the pipeline

Audio capture

Your app pulls a buffer of PCM audio off the input: microphone, line input, or USB capture device.

Rapidly engine

You hand the buffer to a RapidlyEngine instance. The engine removes noise, reverb, and unwanted sounds, then returns a cleaned buffer of the same shape.

Encoder, recorder, or network

You use the cleaned buffer in place of the raw one. Send it to your encoder, write it to a file, or push it to a network stream.

When capture-side is the right choice

Communications apps

Cleaning a phone call, conference call, or VoIP voice before sending.

Voice assistants and dictation

Improving speech-to-text accuracy by feeding cleaner audio.

Live streaming and broadcasting

Cleaning a microphone before encoding for the streaming service.

Voice recording apps

Producing clean recordings without a separate post-processing step.

Embedded devices and hardware

Cleaning audio captured by built-in microphones.

Together with receive-side

Many apps clean both directions. See Receive-side processing for the playback path, and One vs multiple engines for how to set up two engines when both directions matter.

For the API call shapes, see C / C++, Python, or Swift.