More actions
| OpenAI 3DS | |
|---|---|
 | |
| General | |
| Author | MrHuu |
| Type | Other Apps |
| Version | 0.0.1 |
| License | Mixed |
| Last Updated | 2023/08/20 |
| Links | |
| Download | |
| Website | |
| Source | |
OpenAI-3DS features a simple playground for using the OpenAI API. It's basically a chat client with the ability to record your voice and synthesize the response as audio voice as main feature. The response is streamed back, meaning you'll get an almost immediate response after sending your question. Voice synthesizing starts after the first sentences are received, while receiving the rest of the response.
Written in c and using threads and citro2d for graphics allows for smooth scrolling though over 100 message history (including images), while the response is streaming and voice synthesizing. During chat, AI is aware of last 10 questions and responses.
Features
- Models; gpt-3.5 / gpt-4 / davinci-003 / whisper-1 / dall-e.
- Text -> Speech.
- Speech -> Text.
- Speech -> Speech.
- Image generation (256x256).
User guide
Important notes:
- This is a build with core functionality mostly working. But has unfinished and experimental functions.
- Some unhandled errors are to be expected.
- May have unexpected side effects when you start an audible discussion with your 3DS in public.
- Using the OpenAI API will cost you money.
How to setup:
- Enter your API-key on first launch, or press Select in menu to re-enter; OR
- create a api-key.json file in /3ds/OpenAI/
- with the following contents (replaced with your API-Key):
{
"api_key": "sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
}
Screenshots
Known issues
- Crash when using an invalid API-Key.
- Does not work nicely with Citra emulator.
- General coding issues (sorry about that).
- davinci-003 not fully implemented.
Changelog
v0.0.1
- First Release.