Are you using headphones, or literal speakers? For speech recognition in general, a (USB) headset is recommended.
If you are using speakers, your microphone will pick up the sound they produce, which can then either be erroneously be interpreted by speech recognition engine as speech, and/or cause issues recognizing actual speech.
Nahimic can interfere with the ability for the speech recognition engine to use your microphone, and is included with some sound card driver suites.
You'll want to check your sound card settings (including in any extra software from the manufacturer, if present) for microphone filtering options, and try disabling those.
This topic also offers suggestions for recognition issues.
Once your microphone is working correctly, you'll want to train your speech recognition profile, at least three times.
If you do make changes to your microphone configuration (especially when switching to a different one), you'll want to
create a fresh speech recognition profile, rather than combining old training data with new.