FluidAudio

On-device speech recognition via FluidAudio (Apple Core ML) for NativePHP Mobile. Transcribe audio files or stream from the microphone in real time — all on-device, no cloud required.

Note: Android support is a stub only. Full functionality is iOS-only.

Features

• Batch file ASR — download model packs, then transcribe audio files with transcribeFile()
• Microphone streaming — real-time partial transcripts using Parakeet EOU models
• Non-blocking — all calls return immediately with request IDs; results arrive via events
• Multiple model variants — English v2 or multilingual v3 for batch; 160ms/320ms/1280ms latency options for streaming
• Model caching — downloaded models persist on-device for reuse
• JavaScript client — @vikas5914/fluidaudio npm package for Inertia/Vue/React

Installation

composer require vikas5914/fluidaudio
php artisan native:plugin:register vikas5914/fluidaudio

iOS Permissions

The plugin declares NSMicrophoneUsageDescription and audio background mode automatically.