🎵 Realtime Audio

RealtimeAudio is a Flutter package that handles audio recording and playback of data chunks from real-time sources, like OpenAI Realtime, ElevenLabs or HumeAI Voice.

✨ Features

  • 🎤 Audio recording with variable chunk length in milliseconds.
  • 🔊 Audio playback of data chunks.
  • ⏱️ Duration tracking of audio chunks.
  • ⏸️ Pause support.
  • 📊 Volume tracking in dBFS.
  • 🎛️ Voice isolation and other processing on iOS.
  • 📱 iOS audio session handling for max volume.
  • 🤖🍏🍎 Android, iOS, and macOS support.
  • ✂️ Audio response truncation support.

📖 Usage

To use this package, add realtime_audio as a dependency in your pubspec.yaml file.

After initializing the RealtimeAudio object, you can start recording and playing audio chunks by calling start().

Audio will be played back in real-time, as soon as the first chunk is queued with queueChunk().

See the example project for a complete example.

🔒 Permissions

🤖 Android

  • Add Permissions to AndroidManifest.xml:

    <uses-permission android:name="android.permission.RECORD_AUDIO"/>

🍏 iOS

  • Add Permissions to Info.plist: Open your Info.plist file and add the following keys:

    <key>NSMicrophoneUsageDescription</key>
<string>We need access to the microphone to record audio.</string>

🍎 macOS

  • Add Permissions to Info.plist: Open your Info.plist file and add the following keys:

    <key>NSMicrophoneUsageDescription</key>
<string>We need access to the microphone to record audio.</string>

  • Add Entitlements to release and debug macos/*.entitlements: Open your macos/*.entitlements file and add the following:

    <key>com.apple.security.device.audio-input</key>
<true/>

    Can also do this through XCode by selecting the target, then Signing & Capabilities, then checking Audio Input.

Libraries

realtime_audio