Building a voice assistant to control music
smalldata.tech
smalldata.tech
However it's not quite so easy to use - there's a python package that interfaces with it but you would need to know how to grab your own audio sample from the mic and pass it to the library.
I actually managed to get a sample app running using it but I found that the error rate was around 50% and I needed to repeat myself quite a few times.
Not to diminish all the hard work that the Sphinx group has done, but it's only at alpha quality ATM. I think it has potential, but I don't know how fast the development is moving.
You can greatly improve the recognition by using a restricted language model, however.