Nerd Dictation on OpenSUSE, My Installation Tweaks

Date: 19-Nov-2024

Nerd Dictation is a great voice control utility for Linux distros (one of too few)!

Follow the instructions as set out on the repo homepage here,

Download a better VOSK model!

The default install instructions (at the time of writiing) have the user download vosk-model-small-en-us.

My suggestion is simply to download the biggest VOSK model instead:

You might also wish to change the installation path a bit so that it's easier to remember.

I changed mine from nerd-dictation to voice-control.

## Move VOSK model into folder

Now you can move the bigger VOSK model into the folder

Then, simply rename the folder model.

## Enable mic access and verify model is running

This is where I got stuck (my default voice control mic wasn't being recognised).

I needed to install pavucontrol with:

sudo zypper install pavucontrol

and then verify that the correct audio device was enabled (and the non-useful one turned off)

Test and validate

To verify that it's working, you can run the command per the documentation:

./nerd-dictation begin --vosk-model-dir=./model

alt text

Switching models

If you decide that you prefer the performance of a different VOSK model, it's easy to just shuffle them into another folder.

I created one called other-models in my voice control directory and just moved in the original model:

You can of course populate a few different models in the folder. Just give them easy to remember paths:

Then you can just adjust the model path when running the execution command. Just note not to include a trailing slash:

# Shortcut configuration

After finding out which model performs best on your hardware, you can add keyboard shorcuts to start and stop the background dictation process.

## Start background dictation capture

To start

./nerd-dictation begin --vosk-model-dir=./models/small &

To stop

./nerd-dictation begin --vosk-model-dir=./model &

Updating Your Paths For Easy Launching

You can now follow the instructions on the Github repo to move your model

mkdir -p ~/.config/nerd-dictation
mv ./model ~/.config/nerd-dictation

Setting up Bash aliases to start and end voice input

To simplify starting and stopping speech recognition, there are a few approaches that make sense.

One is adding the voice control directory to your path.

Another approach is just creating bash aliases for startingna and stopping voice input.

Let's say I want to set up:

start-voice-control

and:

stop-voice-control

To start and stop voice recognition (respectively) we can use bash aliases for the start and stop commands:

On my OS (change for your paths):

~/voice-control/nerd-dictation begin --vosk-model-dir=~/voice-control/models/large

And:

~/voice-control/nerd-dictation end

So I added:

alias start-voice-control='/~/voice-control/nerd-dictation begin --vosk-model-dir=/~/voice-control/models/large &'

And

alias stop-voice-control=' /home/daniel/voice-control/nerd-dictation end'

Now we can launch and exit the process much more easily

Assign keybindings to shortcuts

Now we can assign keybindings much more easily by just mapping them to the Bash alias commands: