The storm chasing observation tool that turns your voice into documented decisions, observations, and waypoints—with AI feedback to sharpen your chase skills.
OlieJournal is designed for real-time use during storm chases. Speak your observations, log critical decisions, and mark locations—all while keeping your eyes on the storm.
Dictate observations hands-free while driving. OlieJournal transcribes your voice into structured entries instantly—whether you're noting a developing wall cloud, logging a directional decision, or marking a promising intercept location.
Get intelligent insights on your observations to improve your chase decision-making and weather analysis skills. The built-in chatbot analyzes your notes and provides constructive feedback to help you become a more effective storm chaser.
Automatically capture GPS coordinates with each entry. Mark key locations and visualize your chase path on a map. Every entry includes GPS coordinates mapped out with full location context on an interactive map.
Document why you chose specific routes or targets. Build a comprehensive record of decision-making with timestamps, locations, and detailed notes for post-chase analysis and sharing.
Capture real-time conditions, cloud structures, and atmospheric changes while they're happening in the field. Speak naturally—your observations are transcribed accurately so you can focus on the chase.
Built with Flutter for fast, reliable performance in the field—even during data-intensive chase days. Optimized for field conditions with fast transcription, offline capability, and instant GPS logging.
I built OlieJournal as a storm-chasing observation platform that combines a Flutter app with an ASP.NET Core backend and an asynchronous processing pipeline to transcribe audio, generate AI feedback, and create voice responses.
I developed most of the Flutter application with GitHub Copilot, showcasing my ability to ship effectively with AI-assisted tools.
I designed OlieJournal as a multi-tier system built around asynchronous processing:
- Client: Flutter app records audio and captures location metadata.
- API: ASP.NET Core API accepts uploads and orchestrates workflow.
- Queue workers: CLI services process long-running tasks from Azure Service Bus.
- Processing pipeline:
- Google Cloud Speech-to-Text transcribes recordings.
- OpenAI analyzes transcripts and produces coaching feedback.
- Google Text-to-Speech generates spoken responses.
- Storage:
- MySQL stores journal entries, conversations, and metadata.
- Azure Blob Storage stores source audio, converted audio, and backups.
See architecture.txt for the architecture diagram.
-
Capture and upload (Client → API)
- Flutter records audio and attaches location metadata.
- Client uploads payload to the ASP.NET Core API.
-
Persist and enqueue (API)
- API stores raw audio in Azure Blob Storage.
- API writes initial journal metadata to MySQL.
- API publishes a processing message to Azure Service Bus.
-
Transcription stage (Worker)
- Background CLI worker consumes the queue message.
- Worker retrieves blob audio and calls Google Speech-to-Text.
- Transcript and processing status are persisted to MySQL.
-
Analysis stage (Worker)
- Chatbot worker receives transcript-ready work.
- Worker sends transcript + prompt context to OpenAI.
- AI response is saved to conversation/journal tables.
-
Voice synthesis stage (Worker)
- Voiceover worker sends chatbot output to Google Text-to-Speech.
- Generated audio is stored in Blob Storage.
- Final artifact references and completion state are written to MySQL.
-
Operational behavior
- Queue-backed workers decouple API latency from long-running jobs.
- Each stage can be scaled independently based on queue depth.
- Failures are isolated per stage and can be retried via message reprocessing.
- Complete infrastructure setup: infrastructure/README.md
- Complete project installation: install/README.md
Learn more at oliejournal.olievortex.com