-
Notifications
You must be signed in to change notification settings - Fork 1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Lookahead model only recognizes words from word list #35
Comments
I might need to change G.fst construction in sources for that, will look on it a bit later. |
@nshmyrev have you been able to test and repeat this? Where there are a lot of false positives for the words in the lookahead graph |
@yondu22 sure, its kind of reasonable. Add |
thank you [unk] did help a lot, still a lot of false positives but not as much as before |
I was able to get the lookahead model working, but it seems to recognize only words from the wordlist, for example, lets say I pass "one two three" and the person speaks my name is alpha and I can count to one two three, it produces wrong results because the words prior to one two three are also recognized as one two three. I wanted to implement a phrase hint system for example, perform normal asr but put more weight on words passed through a list.
The text was updated successfully, but these errors were encountered: