https://github.com/vgrafe/scarlett

mobile app for voice chat with openai/elevenlabs
https://github.com/vgrafe/scarlett

Last synced: 4 days ago
JSON representation

mobile app for voice chat with openai/elevenlabs

• Host: GitHub
• URL: https://github.com/vgrafe/scarlett
• Owner: vgrafe
• License: mit
• Created: 2023-04-09T16:42:49.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2023-06-18T20:04:38.000Z (over 1 year ago)
• Last Synced: 2024-11-09T15:27:54.671Z (2 months ago)
• Language: TypeScript
• Size: 2.19 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Scarlett

Spoken conversational experience for mobile devices, made with [Expo](https://expo.dev/).

I wanted to fiddle with OpenAI [whisper](https://platform.openai.com/docs/api-reference/audio/create) and [gpt](https://platform.openai.com/docs/api-reference/chat/create), and plug [elevenlabs speech synthesis](https://beta.elevenlabs.io/speech-synthesis) on top of it. This has probably been done way better, but I wanted do put something fun together.

This is a pretty rough POC but good enough to work and see some initial areas of improvement:

- I want to explore a more conversational experience without having to press a button but this is not as easy as it sounds. This would enable the user to cut off the bot as it is answering which could lead to a pretty fun UX for role playing.

- the latency must go. There are a few avenues to do so e.g. running the models on device.