[Feature Request] Read aloud

[nichu42]

I'm wondering if it would be possible for BN to read aloud the replies from Kagi assistant..?

[FaFre]

Yes, it would be possible. You're dreaming about a completely voice-controlled interface, aren't you? :smirk:

[nichu42]

You got me. ;-) In fact, it's just that I sometimes find it easier to communicate by voice for health reasons.

[FaFre]

So I thought about it for a bit and came up with the following:

• The hardest part will be the UI
• For future Assistant communication I will write a native Chat-UI that will enable TTS and integrates the already existing chat archive (+ more features in the future which will be easy to integrate, as we have full control).
• The new UI will be optional and controlled via a setting
• With the new UI, the assistant will run in an background WebView (invisible to the user) which is then "remote controlled" by the new Chat-UI via injected JavaScript
• The injected Js will then hook into the kagi library and channel all incoming messages back the the Chat-UI where they are displayed and further processed (e.g. text-to-speech or auto-archive).
• Chat interactions (like replies) will be evaluated as user interactions via injected JavaScript ("remote control" of WebView)

I think the solution above will be the most robust with best UI options. Injecting TTS buttons or something into the actual Website will be unstable and bad to maintain. Yet the JS Library will be most probably more stable and changes will be less frequent than on the UI side.

I have some ideas regarding the UI but wanna hear yours first.

How do you (or anyone else reading this) imagine interacting with chats in general in the future, and how the UI should look like? When you can contribute some detailed descriptions (or even some sketches), that would be really great!

[nichu42]

I am sorry for the late response. It's definitely not because I lost interest. What you are proposing sounds like an awful lot of work. That was not my intention. I thought that it might be possible to integrate a third-party plug-in or use a system API to just read aloud the output that you get back from Kagi.

Now your suggestions are a completely different story, it seems. I'm not sure if I understood it entirely, so I'll focus on your last question: In general, I'd just like to have a button on my home screen that allows me to ask a question, which is then processed by Kagi assistant (I'd like to choose the default mode in the settings). Preferably, this would happen without the need of any further buttons to be pressed. The app should open at the same time and show the results. Another button should trigger the TTS feature.

[FaFre]

If you have Google apps installed on your phone, you can also use the integrated TTS feature of google for now, which is easily accessible:

I'm thinking a bit ahead of what is possible with the app, usability- & feature wise, so I did a bit of a brainstorming. I plan to make it more useful and direct for mobile research.

Besides that, I got your request with the easy voice input and will be working on something for the next update.