I am creating a project, where i need some specific words to be recognized more often than others. But i don't mean recognizing ONLY these words, i mean recognizing all the words, but just increase the probability that when i say something very similar to key words, a model would most likely recognize that key word, not something close to it. But if the input is not even close to the input word - it will recognize it as it is. Is it possible to achieve?
Sure, you can add more examples to extra.txt and recompile the graph. It will increase probability of your keyword. See https://alphacephei.com/vosk/lm
I am creating a project, where i need some specific words to be recognized more often than others. But i don't mean recognizing ONLY these words, i mean recognizing all the words, but just increase the probability that when i say something very similar to key words, a model would most likely recognize that key word, not something close to it. But if the input is not even close to the input word - it will recognize it as it is. Is it possible to achieve?