This Flutter app allows users to record audio, play back the recorded audio, and receive a transcription by Whisper API of the spoken text displayed in the UI. The app is designed and tested for Android, iOS, iPad, and macOS.
Before you begin, ensure you have met the following requirements:
- You have installed Flutter SDK: Flutter Installation Guide
- You have installed Android Studio or Xcode (for iOS) and set up an emulator or simulator.
- Clone the repository:
git clone https://github.com/ayteksokmen/audio_transcriber.git cd audio_transcriber - Install the dependencies:
flutter pub get
- Generate files (if necessary):
If there is a compilation error, especially on transcription_service.dart file, run the following command to fix
the issue. It will generate code based on the annotations used in transcription_service.dart file.
dart run build_runner build
To run the app on an emulator or physical device, use the following command:
flutter run
The app follows a structured architecture that separates concerns into different layers: the core layer, the data layer, the domain layer, and the presentation layer.
• Bloc: Contains base classes for BLoC state management.
• Params: Contains parameter classes for various operations.
• Resources: Contains classes for managing data state.
• Utils: Contains utility classes and constants.
• Repositories: Implements the repository interfaces defined in the domain layer.
• Services: Contains service classes for interacting with external APIs and functionalities.
• Entities: Defines the data models.
• Repositories: Defines repository interfaces.
• Use Cases: Contains use case classes for various operations.
• Blocs: Contains BLoC classes for managing UI state.
• Screens: Contains the UI screens.
• Widgets: Contains reusable UI components.
The app uses a Firebase Cloud Function to handle audio transcription requests using the Whisper API. This function receives an audio file in Base64 String, processes it, and returns the transcribed text.
Purpose
The Firebase function serves the following purposes:
• Security: Keeps the Whisper API key secure by not exposing it to the client-side.
• Processing: Offloads the transcription processing to the server, reducing the load on the client device.
• Flexibility: Allows easy updates and changes to the transcription logic without needing to update the client app.
Usage
The function is deployed on Firebase and can be invoked from the app to get the transcription of an audio file. For more details on the Firebase function setup and usage, refer to the transcriber.functions README.
The app utilizes several libraries to provide its functionality:
• permission_handler: Handles permission requests.
• record: Handles audio recording.
• flutter_bloc: Provides BLoC state management.
• retrofit: Handles HTTP requests using Retrofit.
• dio: HTTP client used with Retrofit.
• get_it: Dependency injection.
• fluttertoast: Displays toast messages.
• speech_to_text: Provides speech-to-text capabilities.
• audioplayers: Manages audio playback.
• logger: Provides logging functionalities.
• path_provider: Provides paths for file storage.
• firebase_core: Core functionalities for Firebase.
• wave: Displays waveform visualization.