Neovim plugin offer speech to text capability using the Python speech_recognition library.

You can select your favorite engine for speech-to-text: OpenAI whisper, Google, Vosk API etc.. (or any supported speech_recognition engine).

This is basically a tin wrapper around speech_recognition. So, you really want to check out its documentation. Should work OOB with whisper function.

Uses keyboard, which requires admin on Linux. Luckily, only used for detection of <ESC> so without root access you may only lose the escape key functionality.

Using vim-plug:

Plug 'eyalk11/speech-to-text.nvim'

or with Vundle:

Plugin 'eyalk11/speech-to-text.nvim'
PluginInstall

Before running UpdateRemotePlugins, check the Python provider of your VIM installation:

:CheckHealth

To install Python dependencies:

!pip install -r ~/.vim/plugged/speech-to-text.nvim/requirements.txt
:UpdateRemotePlugins

Those are minimal requirements. The speech_recognition library may have more dependencies depending on the platform and/or desired functionality.

It may be possible to install in one line (untested):

Plug 'eyalk11/speech-to-text.nvim', {'do': ':!python -m pip install -r ./requirements.txt \| :UpdateRemotePlugins'}

Installing the plugin adds the following functionality to VIM:

GetVoice(): returns text by speech

:Voice: replaces the selected range with text (by speech). Gets the same parameters as the ConfigureVoice commmand.

You can cancel both voice commands by pressing esc.

:ConfigureVoice [enginename] [paramsdic] - which configures engine parameters (can also be done by yaml config). The enginename and params would be used for the next Voice/GetVoice call.

Mappings are not added by default. Suggested mappings:

nmap <c-L> :Voice<CR>
imap <C-L> <C-R>=GetVoice()<CR>

https://vi.stackexchange.com/users/23502/vivian-de-smedt for helping out with replacing text.