expo-speech-transcriber

On-device speech transcription for Expo apps. Supports iOS (Apple Speech framework) and Android (SpeechRecognizer API).

Features

• 🎯 On-device transcription - Works offline, privacy-focused
• 📱 Cross-platform - iOS 13+ and Android 13+ (API 33)
• 🚀 Multiple APIs - SFSpeechRecognizer (iOS 13+), SpeechAnalyzer (iOS 26+), and Android SpeechRecognizer
• 📦 Easy integration - Auto-configures permissions
• 🔒 Secure - All processing happens on device
• ⚡ Realtime transcription - Get live speech-to-text updates with built-in audio capture
• 📁 File transcription - Transcribe pre-recorded audio files
• 🎤 Buffer-based transcription - Stream audio buffers from external sources for real-time transcription

Installation

npx expo install expo-speech-transcriber expo-audio

Add the plugin to your app.json:

{
  "expo": {
    "plugins": ["expo-audio", "expo-speech-transcriber"]
  }
}

Custom permission message (recommended):

Apple requires a clear purpose string for speech recognition and microphone permissions. Without it, your app may be rejected during App Store review. Provide a descriptive message explaining why your app needs access.

{
  "expo": {
    "plugins": [
      "expo-audio",
      [
        "expo-speech-transcriber",
        {
          "speechRecognitionPermission": "We need speech recognition to transcribe your recordings",
          "microphonePermission": "We need microphone access to record audio for transcription"
        }
      ]
    ]
  }
}

For more details, see Apple's guidelines on requesting access to protected resources.

Note for Android: The plugin automatically adds the RECORD_AUDIO permission to your Android manifest. No additional configuration is required.

Usage

Realtime Transcription

Start transcribing speech in real-time. This does not require expo-audio.

The startActiveListening() function automatically selects the best transcription engine for the current device:

• Android: Uses standard SpeechRecognizer
• iOS 26+: Uses the new SpeechAnalyzer API
• iOS < 26: Falls back to SFSpeechRecognizer

import { Platform } from "react-native";
import * as SpeechTranscriber from "expo-speech-transcriber";

// Request permissions
// Note: requestPermissions() is only needed on iOS
if (Platform.OS === "ios") {
  const speechPermission = await SpeechTranscriber.requestPermissions();
  if (speechPermission !== "authorized") {
    console.log("Speech permission denied");
    return;
  }
}

const micPermission = await SpeechTranscriber.requestMicrophonePermissions();
if (micPermission !== "granted") {
  console.log("Microphone permission denied");
  return;
}

// Use the hook for realtime updates
const { text, isFinal, error, isRecording } =
  SpeechTranscriber.useRealTimeTranscription();

// Start transcription (automatically selects engine)
await SpeechTranscriber.startActiveListening();

// Stop when done
SpeechTranscriber.stopListening();

NOTE: See RecordRealTimeAndTrancribe for an example on how to use Real Time transcription on android.

File Transcription

Transcribe pre-recorded audio files. Our library handles transcription but not recording—use expo-audio to record audio (see expo-audio documentation), or implement your own recording logic with microphone access via requestMicrophonePermissions().

import * as SpeechTranscriber from "expo-speech-transcriber";
import { useAudioRecorder, RecordingPresets } from "expo-audio";

// Record audio with expo-audio
const audioRecorder = useAudioRecorder(RecordingPresets.HIGH_QUALITY);
await audioRecorder.prepareToRecordAsync();
audioRecorder.record();
// ... user speaks ...
await audioRecorder.stop();
const audioUri = audioRecorder.uri;

// Transcribe with SFSpeechRecognizer (preferred)
const text = await SpeechTranscriber.transcribeAudioWithSFRecognizer(audioUri);
console.log("Transcription:", text);

// Or with SpeechAnalyzer if available
if (SpeechTranscriber.isAnalyzerAvailable()) {
  const text = await SpeechTranscriber.transcribeAudioWithAnalyzer(audioUri);
  console.log("Transcription:", text);
}

For custom recording without expo-audio:

// Request microphone permission for your custom recording implementation
const micPermission = await SpeechTranscriber.requestMicrophonePermissions();
// Implement your own audio recording logic here to save a file
// Then transcribe the resulting audio file URI

Buffer-Based Transcription

Stream audio buffers directly to the transcriber for real-time processing. This is ideal for integrating with audio processing libraries like react-native-audio-api.

import * as SpeechTranscriber from "expo-speech-transcriber";
import { AudioManager, AudioRecorder } from "react-native-audio-api";

// Set up audio recorder
const recorder = new AudioRecorder({
  sampleRate: 16000,
  bufferLengthInSamples: 1600,
});

AudioManager.setAudioSessionOptions({
  iosCategory: "playAndRecord",
  iosMode: "spokenAudio",
  iosOptions: ["allowBluetooth", "defaultToSpeaker"],
});

// Request permissions
const speechPermission = await SpeechTranscriber.requestPermissions();
const micPermission = await AudioManager.requestRecordingPermissions();

// Stream audio buffers to transcriber
recorder.onAudioReady(({ buffer }) => {
  const channelData = buffer.getChannelData(0);
  SpeechTranscriber.realtimeBufferTranscribe(
    channelData, // Float32Array or number[]
    16000, // sample rate
  );
});

// Use the hook to get transcription updates
const { text, isFinal, error } = SpeechTranscriber.useRealTimeTranscription();

// Start streaming
recorder.start();

// Stop when done
recorder.stop();
SpeechTranscriber.stopBufferTranscription();

See the BufferTranscriptionExample for a complete implementation.

Integration with Expo Audio Studio

You can easily pipe audio from expo-audio-studio directly to the transcriber. This works cross-platform and uses the optimal transcription engine (SpeechAnalyzer on iOS 26+, SFSpeechRecognizer on older iOS, etc).

import { AudioRecorder } from 'expo-audio-studio';
import * as SpeechTranscriber from 'expo-speech-transcriber';

// ... inside your component or function ...

// 1. Request permissions
const permission = await SpeechTranscriber.requestPermissions(); 
// ... handle permission ...

// 2. Start recording with expo-audio-studio
await AudioRecorder.startRecording({
    sampleRate: 16000,
    channels: 1,
    encoding: 'pcm_16bit', // Use standard 16-bit PCM (default for Audio Studio)
    bufferDurationSeconds: 0.1, // 100ms updates
    output: {
        primary: { enabled: false } // We don't need a file, just the stream
    },
    onAudioStream: async (data) => {
        // data.data is the Base64 string of the audio buffer
        // expo-speech-transcriber's `realtimeBufferTranscribeBase64` handles
        // the Base64 decoding and Int16 -> Float32 conversion natively for performance.

        if (data.data) {
             await SpeechTranscriber.realtimeBufferTranscribeBase64(data.data, 16000);
        }
    }
});

// 3. Listen for results (same hook as always)
const { text } = SpeechTranscriber.useRealTimeTranscription();

// 4. Stop
await AudioRecorder.stopRecording();
SpeechTranscriber.stopBufferTranscription();

API Reference

requestPermissions()

Request speech recognition permission.

Platform: iOS only. On Android, speech recognition permission is handled through requestMicrophonePermissions().

Returns: Promise<PermissionTypes> - One of: 'authorized', 'denied', 'restricted', or 'notDetermined'

Example:

import { Platform } from "react-native";

if (Platform.OS === "ios") {
  const status = await SpeechTranscriber.requestPermissions();
}

requestMicrophonePermissions()

Request microphone permission.

Returns: Promise<MicrophonePermissionTypes> - One of: 'granted' or 'denied'

Example:

const status = await SpeechTranscriber.requestMicrophonePermissions();

startActiveListening()

Start real-time speech transcription using the optimal engine for the device (Android, iOS SpeechAnalyzer, or iOS SFSpeechRecognizer). Listen for events via useRealTimeTranscription hook.

Returns: Promise<void>

Example:

await SpeechTranscriber.startActiveListening();

recordRealTimeAndTranscribe()

Start real-time speech transcription. Listen for events via useRealTimeTranscription hook.

Returns: Promise<void>

Example:

await SpeechTranscriber.recordRealTimeAndTranscribe();

stopListening()

Stop real-time transcription.

Returns: void

Example:

SpeechTranscriber.stopListening();

isRecording()

Check if real-time transcription is currently recording.

Returns: boolean

Example:

const recording = SpeechTranscriber.isRecording();

transcribeAudioWithSFRecognizer(audioFilePath: string)

Transcribe audio from a pre-recorded file using SFSpeechRecognizer. I prefer this API for its reliability.

Platform: iOS only

Requires: iOS 13+, pre-recorded audio file URI (record with expo-audio or your own implementation)

Returns: Promise<string> - Transcribed text

Example:

const transcription = await SpeechTranscriber.transcribeAudioWithSFRecognizer(
  "file://path/to/audio.m4a"
);

transcribeAudioWithAnalyzer(audioFilePath: string)

Transcribe audio from a pre-recorded file using SpeechAnalyzer.

Platform: iOS only

Requires: iOS 26+, pre-recorded audio file URI (record with expo-audio or your own implementation)

Returns: Promise<string> - Transcribed text

Example:

const transcription = await SpeechTranscriber.transcribeAudioWithAnalyzer(
  "file://path/to/audio.m4a"
);

isAnalyzerAvailable()

Check if SpeechAnalyzer API is available.

Platform: iOS only. Always returns false on Android.

Returns: boolean - true if iOS 26+, false otherwise

Example:

if (SpeechTranscriber.isAnalyzerAvailable()) {
  // Use SpeechAnalyzer
}

useRealTimeTranscription()

React hook for real-time transcription state.

Returns: { text: string, isFinal: boolean, error: string | null, isRecording: boolean }

Example:

const { text, isFinal, error, isRecording } =
  SpeechTranscriber.useRealTimeTranscription();

realtimeBufferTranscribe(buffer, sampleRate)

Stream audio buffers for real-time transcription. Ideal for integration with audio processing libraries.

Parameters:

• buffer: Float32Array | number[] - Audio samples
• sampleRate: number - Sample rate in Hz (e.g., 16000)

NOTE We currently support transcription for mono audio only. Natively, the channel is set to 1.

Returns: Promise<void>

Example:

const audioBuffer = new Float32Array([...]);
await SpeechTranscriber.realtimeBufferTranscribe(audioBuffer, 16000);

stopBufferTranscription()

Stop buffer-based transcription and clean up resources.

Returns: void

Example:

SpeechTranscriber.stopBufferTranscription();

Language Configuration

setLanguage(localeCode: string)

Set the language for speech transcription. Must be called before starting transcription.

Platform: iOS and Android

Parameters:

• localeCode: string - BCP-47 locale code (e.g., 'es-ES', 'fr-FR', 'ar-SA', 'zh-CN')

Returns: Promise<void>

Example:

// Set to Spanish (Spain)
await SpeechTranscriber.setLanguage('es-ES');

// Set to Arabic (Saudi Arabia)
await SpeechTranscriber.setLanguage('ar-SA');

// Set to Chinese (China)
await SpeechTranscriber.setLanguage('zh-CN');

// Then start transcription
await SpeechTranscriber.recordRealTimeAndTranscribe();

Common Locale Codes:

| Language | Locale Code | |----------|-------------| | English (US) | en-US | | English (UK) | en-GB | | Spanish (Spain) | es-ES | | Spanish (Mexico) | es-MX | | French (France) | fr-FR | | German (Germany) | de-DE | | Italian (Italy) | it-IT | | Portuguese (Brazil) | pt-BR | | Russian (Russia) | ru-RU | | Chinese (Simplified) | zh-CN | | Chinese (Traditional) | zh-TW | | Japanese | ja-JP | | Korean | ko-KR | | Arabic (Saudi Arabia) | ar-SA | | Hindi (India) | hi-IN | | Dutch (Netherlands) | nl-NL | | Polish (Poland) | pl-PL | | Turkish (Turkey) | tr-TR | | Vietnamese | vi-VN |

getAvailableLanguages()

Get a list of all languages supported on the current device.

Platform: iOS and Android

Returns: Promise<string[]> - Array of locale codes

Example:

const languages = await SpeechTranscriber.getAvailableLanguages();
console.log('Available languages:', languages);
// Output: ['en-US', 'es-ES', 'fr-FR', 'de-DE', ...]

getCurrentLanguage()

Get the currently selected language for transcription.

Platform: iOS and Android

Returns: Promise<string> - Current locale code

Example:

const current = await SpeechTranscriber.getCurrentLanguage();
console.log('Current language:', current);
// Output: 'en-US'

isLanguageAvailable(localeCode: string)

Check if a specific language is available on the current device.

Platform: iOS and Android

Parameters:

• localeCode: string - BCP-47 locale code to check

Returns: Promise<boolean> - true if available, false otherwise

Example:

const isSpanishAvailable = await SpeechTranscriber.isLanguageAvailable('es-ES');
if (isSpanishAvailable) {
  await SpeechTranscriber.setLanguage('es-ES');
  console.log('Spanish is available!');
} else {
  console.log('Spanish is not supported on this device');
}

Complete Multi-Language Example

import * as SpeechTranscriber from 'expo-speech-transcriber';
import { Platform } from 'react-native';

async function transcribeInSpanish() {
  // Request permissions
  if (Platform.OS === 'ios') {
    const speechPermission = await SpeechTranscriber.requestPermissions();
    if (speechPermission !== 'authorized') {
      console.log('Speech permission denied');
      return;
    }
  }

  const micPermission = await SpeechTranscriber.requestMicrophonePermissions();
  if (micPermission !== 'granted') {
    console.log('Microphone permission denied');
    return;
  }

  // Check if Spanish is available
  const isSpanishAvailable = await SpeechTranscriber.isLanguageAvailable('es-ES');
  if (!isSpanishAvailable) {
    console.log('Spanish not supported on this device');
    
    // Show available languages
    const languages = await SpeechTranscriber.getAvailableLanguages();
    console.log('Available languages:', languages);
    return;
  }

  // Set language to Spanish
  await SpeechTranscriber.setLanguage('es-ES');
  console.log('Language set to Spanish');

  // Start transcription
  await SpeechTranscriber.recordRealTimeAndTranscribe();
  console.log('Recording in Spanish...');
}

// For file transcription with different language
async function transcribeFileInFrench(audioUri: string) {
  await SpeechTranscriber.setLanguage('fr-FR');
  const text = await SpeechTranscriber.transcribeAudioWithSFRecognizer(audioUri);
  console.log('French transcription:', text);
}

Example

See the example app for a complete implementation demonstrating all APIs.

Requirements

iOS

• iOS 13.0+
• Expo SDK 52+
• Development build (Expo Go not supported - why?)

Android

• Android 13+ (API level 33)
• Expo SDK 52+
• Development build (Expo Go not supported)

Limitations

• Language availability - Supported languages vary by device and OS version. Use getAvailableLanguages() to check what's supported on the current device
• File size - Best for short recordings (< 1 minute)
• Recording not included - Real-time transcription captures audio internally; file transcription requires pre-recorded audio files (use expo-audio or implement your own recording with requestMicrophonePermissions())
• Android file transcription - File-based transcription (transcribeAudioWithSFRecognizer, transcribeAudioWithAnalyzer) is iOS only. Android supports real-time transcription
• Android API level - Android requires API level 33+ (Android 13)

License

MIT

Contributing

Contributions welcome! Please open an issue or PR on GitHub.

Author

Dave Mkpa Eke - GitHub | X