I would like my speech to text api to recognize speech, but only output words from a list I provide it. This is because in my app I only want to recognize certain commands, and not regular speech, but the commands sometimes get mistaken for regular speech and then the output is different than the command string, ruining the app. Is there a way with any existing speech-to-text API/package to use only custom vocabulary as output and nothing else?
I tried the google api and the web speech api but it only favors custom vocabulary, it could also output other words.
Thanks
based on my understanding speech-text apis or packages recognise the entire speech. One possible way it to do text comparison. comparing the speech text with your custom commands.