Bring Ted to life
Next Steps
Transcription:
- add automatic transcription (2)
- dynamic audio recording (ends after 3s of no speech found) (3)
- use llm context window/memary to store transcription & responses for context (clear graph for every new conversation) (9)
Model:
- change to llama 3 8B (1)
Server:
- Openvoice added to ec2 instance (5)
- benchmark ollama vs server response times (8)
Speech:
- openvoice over Ted's voice (4)
- speaker integration (6)
Design:
-
redesign & print 3d case (7)
-
Build Teddy (insert mcu & speaker) (10)
-
Documentation (11)
Done:
- CConnect mcu & battery (hardware)
- Gather data (software)
- transcribe using mcu
- host llama 2 on ec2 instance
