GPTandem

The idea is to leverage AI capabilities to create an artificial tandem partner.
The application is built using Java and Swing.

Flow

1. Record voice input
2. Send audio to Google (speech-to-text) in order to obtain a transcript of what the user said
3. Use the transcript to query OpenAI APIs (chatGPT 3.5) to obtain a textual response
4. Use Google TTS APIs to obtain an audio version of ChatGPT response
5. Repeat

Note

In order to use the program you must provide your own Google and OpenAI tokens as the services are paid to use. Google uses a JSON file, and the path to the file must be provided using the env. var. GOOGLE_APPLICATION_CREDENTIALS, for openAI the env. var. to set is OPENAI_KEY