search

ccoreilly / vosk-browser

A speech recognition library running in the browser thanks to a WebAssembly build of Vosk
Apache License 2.0
364 stars 60 forks source link

How to reduce delay between results? #43

Closed bbergeron0 closed 2 years ago

bbergeron0 commented 2 years ago

Hi, I’ve noticed Vosk wait a few split seconds after the user is done talking before emitting a result, unlike partial results which get fired continuously, but aren’t as reliable. Since our application only respond to short, single-word commands, we’d like to reduce Vosk’s result “de-bounce” time to make our application feels more responsive. Do you have any suggestions?

ccoreilly commented 2 years ago

Hi @3NdR1m ! You can edit the endpoint rules in model.conf of the model you are using, to define when online decoding should stop (i.e. a final result should be sent). Here is the reference.

bbergeron0 commented 2 years ago

Setting --endpoint.rule1.min-trailing-silence= through --endpoint.rule4.min-trailing-silence= seems to works. Thanks for your time.

actor10 commented 1 year ago

Setting --endpoint.rule1.min-trailing-silence= through --endpoint.rule4.min-trailing-silence= seems to works. Thanks for your time.

What values did you use? (My program uses commands of just a word or two.). Henry