Skip to main content
whisper.rn Hero Light

​
What is whisper.rn?

whisper.rn is a React Native binding for whisper.cpp, bringing OpenAI’s powerful Whisper automatic speech recognition (ASR) model to iOS and Android devices with high-performance native inference.

On-Device Processing

All transcription happens locally on the device - no cloud APIs, no internet required, complete privacy

Production Ready

Battle-tested in production apps with optimized performance for mobile devices

Hardware Acceleration

Supports Metal GPU on iOS and Core ML for encoder acceleration

Realtime Transcription

Built-in support for live microphone transcription with Voice Activity Detection

​
Key Features

  • File Transcription: Process audio files (WAV, base64 encoded audio)
  • Realtime Transcription: Live microphone input with VAD auto-slicing
  • Data Transcription: Direct PCM data processing via JSI (zero-copy ArrayBuffer)
Detect speech segments using the Silero VAD model with configurable presets:
  • default, sensitive, very-sensitive
  • conservative, very-conservative
  • continuous, meeting, noisy
iOS:
  • Core ML encoder acceleration (iOS 15.0+)
  • Metal GPU acceleration
  • Pre-built xcframework for faster builds
  • Flash Attention support
Android:
  • Optimized NDK builds
  • 16KB page size support (Android 15+)
  • ARM NEON optimizations
All Whisper model sizes supported:
  • tiny, tiny.en (39-75 MB)
  • base, base.en (142-148 MB)
  • small, small.en (466-488 MB)
  • medium, medium.en (1.5 GB)
  • large-v1, large-v2, large-v3 (2.9 GB)
Plus quantized models for reduced size and faster inference.

​
Performance

On an iPhone 13 Pro Max, the tiny.en model with Core ML processes a 30-second audio clip in approximately 800ms in release mode.
Actual performance varies by:
  • Device CPU/GPU capabilities
  • Model size (tiny/base/small/medium/large)
  • Audio duration and complexity
  • Whether Core ML/GPU acceleration is enabled

​
Quick Navigation

Installation

Get whisper.rn installed in your React Native app with platform-specific setup

Quick Start

Transcribe your first audio file in under 5 minutes

API Reference

Complete API documentation for all features

Examples

Real-world examples and use cases

​
Use Cases

  • Voice Notes: Transcribe voice memos and notes
  • Accessibility: Real-time captions for video content
  • Meeting Transcription: Record and transcribe meetings
  • Language Learning: Transcribe practice conversations
  • Voice Commands: Build custom voice-controlled interfaces
  • Subtitles: Generate subtitles for video content

​
Community

whisper.rn is open source and built by BRICKS.

GitHub Repository

Star, fork, and contribute on GitHub

Report Issues

Found a bug? Let us know
Ready to get started? Head over to Installation to set up whisper.rn in your project.

Build docs developers (and LLMs) love

Get started for freeTalk to us