Install Wfloat

Use Python 3.9 or newer:

python -m pip install wfloat

Platform wheels bundle the native runtime; NumPy is installed as a dependency. If pip cannot find a wheel for your platform, a source installation requires a native build.

Optional Pydantic and dataclass schemas:

python -m pip install 'wfloat[schemas]'

The API is synchronous. Microphone capture and audio playback belong to your application.

Next: choose a model and follow its Python example. See model management for caching and cleanup.

Install Wfloat

In your React Native application:

npm install @wfloat/react-native-wfloat
cd ios && pod install

Rebuild the native app after installing. Android uses autolinking. Expo Go cannot load Wfloat's native module; use a native development build. The tested development baseline is React Native 0.76.5 with Hermes and the New Architecture.

For microphone input, add NSMicrophoneUsageDescription to the iOS app's Info.plist and android.permission.RECORD_AUDIO to the Android app's manifest. Wfloat requests microphone permission when capture starts. Playback alone does not require microphone permission.

Next: choose a model and follow its React Native example. See model management for downloads and cleanup.

Install Wfloat

In your web application:

npm install @wfloat/wfloat-web

Import Wfloat in browser-side code. Use a bundler with module-worker support, such as Vite; do not load models during server rendering. Microphone capture requires HTTPS or localhost. Start audio playback or microphone capture from a user interaction.

Next: choose a model and follow its Web example. See model management for download progress and cleanup.