Project Talk to ChatGPT - actual vocal discussion with an AI using voice recognition and text-to-speech in Chrome

[Costello]

Talk-to-ChatGPT is a Google Chrome Extension that allows users to interact with the ChatGPT AI using their voice, rather than just by typing. With this tool, users can speak to the AI and receive spoken responses, making the interaction feel more natural and conversational. This could be useful in a variety of settings where it would be helpful to have a more human-like interaction with an AI. (Note: the AI itself wrote this sentence after I explained it what my code does).



Download Chrome extension here: https://chrome.google.com/webstore/detail/talk-to-chatgpt/hodadfhfagpiemkeoliaelelfbboamlk
Full project source code, FAQ, documentation: https://github.com/C-Nedelcu/talk-to-chatgpt

I will be posting updates in this thread when new versions are made available

 

[sombrerosonic]

Now i can talk to a robot peter griffin on a instance

 

[Costello]

OK the project is up on Github now.

I will be working on the config and demo videos later today

 

[SylverReZ]

Didn't know that Costello was a programmer, enjoy your follow from me on GitHub.

 

[Costello]

I have just published an update : the bot’s response is now read as it is streamed, you no longer have to wait until you receive the complete reply.

This makes for a much more natural conversation ! Much improved.

I will record new demo videos by Monday if I find the time.

 

[MokTan]

Thx. everything works fine but :
1. if you change tab ChatGPT starts listening again no matter of : PAUSE command. Please "fix" this.
2. is there any option with NO VOICE response? Just me asking question aloud and ChatGPT response only writing response with no voice

Post automatically merged: Dec 18, 2022

II. After you close browser settings are not SAVED.

 

[Costello]

v1.5 available



It now comes as a Chrome extension. The extension is under review at the moment, it should reach the web store in the coming days.

Meanwhile if you want to load it in developer mode, you can download it below.
Then, how to load an extension in developer mode: https://webkul.com/blog/how-to-install-the-unpacked-extension-in-chrome/

 

[Costello]

The Chrome extension is now available on the web store :
https://chrome.google.com/webstore/detail/talk-to-chatgpt/hodadfhfagpiemkeoliaelelfbboamlk
Gimme some good reviews pls!

Some dude published his own extension using my code, without my authorization, and without even crediting me or linking to my project. He even called his extension the same as mine. I emailed him about it but he didnt reply.

 

[SylverReZ]

Found this rather interesting that came up.

 

[Costello]

V1.6 now available
Please note: extension update is still under review, playstore still lists 1.5 at the moment, should be good in a couple of days. Meanwhile you can download the zip and sideload it from the github repo.

Changes are:
- NEW: an 'autosend' option has been added (enabled by default to maintain legacy behavior), it can be disabled to send messages manually
- NEW: 'send word' option to use when autosend is disabled. Just say the word/sequence of words to send the message. Default is set to: 'send message now'
- NEW: a help popup for the extension, which shows when you click the extension icon in the browser's extension bar
- UPDATED: default voice speed set to 1.0 to avoid scrambling
- UPDATED: support for japanese/chinese punctuation
- FIXED: skip button didn't work properly after a recent update
- FIXED: javascript error message when opening ChatGPT with extension for the first time

 

[JackTheHorse]

No Firefox version?

 

[Costello]

No Firefox version?

Unfortunately I tried it on Firefox (using the javascript code alone, not as an extension) but the voice/speech APIs don't work properly
this isn't surprising, because both text-to-speech and speech-to-text require a bit of server-side computation, Google can afford that, but Firefox doesnt have a multibillion dollar company backing it up
I'll be looking out for Firefox updates and we'll see if they eventually properly implement the APIs

 

[tabzer]

This is amazing and underappreciated. I turned my TV into a genius robot. I can communicate with it in English and Japanese. Life is going warp-speed.

 

[Costello]

This is amazing and underappreciated. I turned my TV into a genius robot. I can communicate with it in English and Japanese. Life is going warp-speed.

thanks!
have you tried the recent v2.6 which supports Elevenlabs API? the voices are amazing

 

[Sm3bul0ck]

Crazy to see how fast this is all being developed