Pinecone
Download Audio
Transcribe with
Whisper
-Compute Embeddings
Pinecone
Save Embeddings into pinecone
Compute Question Embedding
Complete Chat with Question +
provided transcriptions and
knowledge
Pinecone
- Get relevant transcriptions
yt-whisper
localhost:8503
dockerbot
Localhost:8504
Docker compose
Explanation
The image describes a workflow involving audio transcription and question answering. The process starts with downloading audio and transcribing it using Whisper, computing embeddings using Pinecone, and saving them. Then, it computes question embeddings, uses provided transcriptions and knowledge to answer questions, and retrieves relevant transcriptions.