ChatGPT integration (with audio output)

[danic85]

There has been a discussion recently around integrating ChatGPT in the following way:

1. Existing speech recognition publishes text in speech event.
2. New ChatGPT module subscribes to speech event and publishes text to ChatGPT via API.
3. Response from ChatGPT is published in tts event (output).
4. TTS module converts TTS and outputs over audio.

There is currently a limitation here as the audio output has been disabled to facilitate Neopixels.
If that were resolved, then the v3 branch already includes a text-to-speech (TTS) module that is configured to output anything published to the topic tts:

pub.sendMessage('tts', msg='I am awake')

Therefore the new ChatGPT model could subscribe to speech and publish tts. No further integration would be required.

Reference:
Creating a module

[WhoIsMrSentry]

A little footnote from what I researched with Dan.As you will use it as sound output do not use raspberry pi.
Raspberry pi neopixel leds incompatible with it you need to turn off the sound for it to work properly use cards like jetson nano.

[danic85]

Thanks @SentryCoderDev . I have created another idea to discuss the audio output on the Pi:

#19

We can continue discussing the ChatGPT use case here.

Feel free to create a PR if you create the new module as it would be a great addition. Don't include any private keys or access tokens in the repository, obviously.

[WhoIsMrSentry]

Hi, after months of research, I found a chatgpt site that can be used without a free API and wrote sample code and created a new PR.

#55

[danic85]

Nice! I'll take a look when I'm back at a computer 👏

[danic85]

Integration merged to develop https://github.com/makerforgetech/modular-biped/blob/develop/docs/ChatGPT.md