← Back to Projects
leptons1618@anishgiri.dev:~/projects/echonode cat project.toml
EchoNode
A local-first chat client for Ollama — switch between five open models mid-conversation, talk to it or type to it, and keep every token on the machine it runs on.
- category
- AI / ML
- year
- 2023
- status
- Stable
- stack
- 8 tools
- highlights
- 5
01Highlights
- Five Ollama models — Llama 2, Mixtral, Mistral, Phi-2 and Vicuna — switchable mid-conversation without restarting anything.
- Voice both ways: `speech_recognition` for capture, `pyttsx3` for synthesis, so a turn can be spoken rather than typed.
- Generation is pushed onto a `ThreadPoolExecutor` (`MAX_WORKERS`, default 5) so the interface stays responsive while a 7B model is still thinking.
- Every knob is an environment variable — Ollama base URL, host, port, worker count, history depth and response cache size.
- Explicit `psutil` and `gc` handling around model residency, which is the difference between a chat app and an out-of-memory one.
02Built with
- 01Python
- 02NiceGUI
- 03Ollama
- 04PyTorch
- 05aiohttp
- 06psutil
- 07SpeechRecognition
- 08pyttsx3