Getting started

1. Add the package.

   flutter pub add transcribe_cpp

   It runs on iOS 16+ and Android 7+ (API 24, 64-bit). See Platforms for the details.

2. Let the app use the microphone and the network.

   The package declares no permissions of its own. Your app asks for the microphone, and needs the network once, to download a model.

   iOS

   In ios/Runner/Info.plist, say why you need the microphone. iOS shows this text in the permission prompt:

   <key>NSMicrophoneUsageDescription</key>
   <string>Your speech is transcribed on this device.</string>

   Android

   In android/app/src/main/AndroidManifest.xml, inside <manifest>:

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

   Debug builds add INTERNET automatically, so if you forget it, you will only find out in a release build.

3. Download a model.

   Models are not bundled with your app. ModelManager downloads the one you ask for, checks it, and keeps it for next time:

   final models = await ModelManager.open();
   final model = await models.ensure(
     const ModelId('parakeet-tdt_ctc-110m-Q4_K_M'), // 90 MB, English
   );

   The first call downloads; after that it returns immediately. See Models for choosing one and showing progress.

4. Transcribe.

   final session = await TranscribeSession.open(model);
   
   await session.start(); // asks for the microphone the first time
   // ...the user speaks; each sentence is transcribed when they pause...
   await session.stop();  // waits for the last sentence
   
   await session.dispose(); // when you no longer need the model

   Show the text with a ValueListenableBuilder. It updates as the user speaks:

   ValueListenableBuilder(
     valueListenable: session.text,
     builder: (context, text, _) => Text(text.full),
   )

Tip

Opening a session loads the model, which takes a moment. Open it once, and call start() and stop() as often as you like.

Next steps

• Transcribing from the microphone: live text, sentences, interruptions and the permission prompt.
• Bringing your own audio: if your app already records audio.
• Listening in the background: keep transcribing after the user leaves the app.
• The example app: a complete app with a model picker.