Did you read the article..? It talks about how speech is one of the most complicated actions / combination of motions the brain performs, and that they're reading motor I/O from multiple regions and functions (breath, lips etc) in order to recreate that into words.

We've had the technology for GUI manipulation through brain-computer interfaces for a while now, controlling a cursor on screen was accomplished long before any form of working speech recognition.

> You could give paralyzed people the ability to use computer again through the cmd console.

Why would they be limited to a cmd console interface...? It's not like the I/O of brain-computer interfaces is limited to text