Inspiration
I have always wanted to build an audio journaling app because I have observered there are times when I want to capture some ideas or express some thoughts that would take too long to write out. Additionally the LLM analysis of captured entries could be used to derive insights for the user.
What it does
This is an app that allows users to generate journal entries through speech. Users audio entries are converted into text and stored in a database. Additionally the generated text is passed to an LLM to help the user gain some insights into what they spoke about.
How we built it
We used Flask and PostgreSQL to build the backend. Android Studio in Java was used to implement the frontend.
Challenges we ran into
Some challenges we faced was the implementation of the transcription and analysis. By far the largest challenge was the integration of the app and the api. We had to learn Celery in order to queue the text transcription and analysis in the background. The frontend had to periodically poll the API until the both of these tasks were completed.
Accomplishments that we're proud of
The text analysis feature was not built and we had not properly tested or integrated the frontend and backend 2 hours before the end of the hackthon.
What we learned
We learned how to integrate llms and ai models into an API. We learned how to asynchronously run background tasks with Celery.
What's next for Audio Journaling
- Improve the analysis of the text
- Build frontend screens for listing, searching, and editing journal entries
Built With
Flask Android Studio Deepseek OpenAI Whisper
Log in or sign up for Devpost to join the conversation.