Web app enabling users to record or upload audio files, utilizing OpenAI API (Whisper, GPT-4) and custom agents/ tools with LangChain to generate transcriptions, summaries, fact checks, sentiment analysis, and text metrics. Additionally, users can interact with a GPT4 chatbot about their transcriptions. Data is stored in an SQLite DB for user authentication and later retrieval. Data is also embedded into a Pinecone vector DB, for LLM purposes.
- Programming Languages: Python, HTML, CSS
- Web Framework: Streamlit
- Audio Transcription: OpenAI API (Whisper)
- LLM: OpenAI API (GPT-4), LangChain
- Text Analysis: NLTK
- Vector DB: Pinecone
- Relational DB and User Auth: SQLite
App V12: Implemented Sidebar for LLM Temperature and Model Selection & Much Improved Code Modularity
App V10: Implemented a Pinecone DB utilizing OpenAI Embedding, and and implemented as optional reference tool for LangChain Zero Shot React Description agent.
From JRE - Neil DeGrasse Tyson - https://www.youtube.com/shorts/y9RowvoU43Q
