Chat with GPT is an open-source TypeScript/React app that integrates OpenAI ChatGPT with ElevenLabs for voice and Whisper for speech recognition, self-hostable via Docker. It supports system prompts, chat search, markdown, and per-user API key handling.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Chat with GPT is an open-source unofficial ChatGPT app that connects ChatGPT with ElevenLabs to provide a realistic human voice. It is built with TypeScript and React and can be run self-hosted via Docker. It supports features like system prompt customization, markdown rendering, chat session search, and per-message editing/regeneration.
How it works
The app uses OpenAI for chat responses, ElevenLabs for text-to-speech, and OpenAI Whisper for speech recognition. Users supply API keys for OpenAI and ElevenLabs, which are stored only on the user’s device when configured in the app. Voice playback and speech input are provided through these integrations.
Getting started
Install/run via Docker with:
docker run -v $(pwd)/data:/app/data -p 3000:3000 ghcr.io/cogentapps/chat-with-gpt:release
Then navigate to http://localhost:3000. To store API keys on the server, create a data/config.yaml with:
services:
openai:
apiKey: (your api key)
elevenlabs:
apiKey: (your api key)
Restart the server after adding the config.
Notes on keys
- OpenAI API key usage is billed at a pay-as-you-go rate. Keys are stored only on the user’s device when entered through the app.
- ElevenLabs API key is stored only on the user’s device.
Recent releases
- The latest releases section indicates no releases are listed in the repository (latest 0).
Traction
- Stars: 2352 (as provided). Forks: 475. Open issues: 97. Language: TypeScript. License: MIT.
Caveats
- License: MIT. Created: 2023-03-06. Last push: 2024-05-24. No explicit latest release tag in the Releases section.
Additional details
- Features include fast response times, chat session search, system prompt customization, configurable creativity via Temperature, voice via ElevenLabs or browser TTS, Whisper-based speech recognition, shareable chat sessions, copy/paste, message editing/regeneration, full markdown support, and usage-based API cost.