Voice to text way

[vesteve58]

I need to know if anybody knows the way in Elive to write in libreoffice text.

Thanks

Vicente

[triantares]

There is no specific Elive way other than the options available to all distros.

That said, running ‘speech-to-text’ locally (i.e on your machine) requires a hefty CPU/GPU so you might want to look at online options.

I’d opt for using your phone with ‘gboard’ and share the generated text to your machine.

Type with your voice - Android - Gboard Help

AND (if you’re feeling adventurous) you could build your own using a RaspberryPi.

Speech Recognition on raspberry Pi Zero

[Thanatermesis]

@triantares

There is no specific Elive way other than the options available to all distros.

Yes there is

But is an Elive Premium feature (patreon members) and requires also an AI API key, I use it all the time and is amazing, so is extremely handy to just speak what you want to type and it will be typed, especially when it recognizes correctly so you don’t have to correct the text, when the voice is recognized with AI it includes correctly all dots, commas, correctly recognizing the words, it even recognizes correctly the word “Elive” or my accent.

In short, you can install it with running:

~ ❯❯❯ elive-premium  ai

And then you can add to your ~/.e16/bindings.cfg :

KeyDown    A    space                    exec elive-assistant --type listen

Reload E with “ctrl + alt + end” and you can use it by pressing “alt + space” (or other wished combo), start talking, and press the hotkey again to process your speech to start typing it… it simulates keyboard presses so it works in any application, GUI or terminals

This feature is not even listed on the website of Elive, my future plans were to create a widget (in E27+) for it for easier / faster use

Important notes:

• elive-assistant actually uses the OPENAI service, I know that there’s other options but they are not implemented at the moment
• this provider, just like many others, tracks you, if you want a service that doesn’t tracks you the only way is to install a speech recognizer locally, or in an external machine in your network, like @triantares proposed with the raspberry machine
• voice recognition is cpu/gpu expensive, or more like proportional… light options doesnt normally recognizes voice in a good quality, AI based ones recognizes it better but is more expensive, but there’s projects that can install it locally for example based on light versions of whisper (i dont remember now a link that I have seen long time ago)
• there’s also apps (with a widget?) that you can install like dsnote or nerd-dictation, i didn’t tried them, im sure there’s more

There’s for sure good projects that we could install easily and light in Elive but we should experiment with them

If you experiment with them, maybe in a tests machine which you can easily reinstall, will be good to know the good options that we could maybe provide on Elive

[triantares]

Actually

I’d be tempted to create a widget making it easy to connect to and use your phone likes a sort of dicta-phone as the easiest solution.

Especially to keep Elive friendly for older hardware.

Also consider that:

A fast and easy way to locally connect and share files (on the fly) would be a “ must have” for Android based phones.