alphacep / alphacep/vosk-api

Vosk grammar question

Open
#1,934 3 comments 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
15.1k
Forks
1.8k
PR merge metrics
No merged PRs in 30d

Description

I'm working on commands triggered system using the `vosk-model-small-en-us-0.15` model.

I have created a grammar with all the keywords that I need, and I have also added `[unk]` because without it the model will transcribe random words as commands.

I'm currently facing 2 issues:

1. I'm trying to capture 3 consecutive numbers, but hasn't been very accurate especially with the number "two" it would occasionally not get captured for example:

- "two two two" would sometimes get transcribed as "two two"
- "one two one" would sometimes get transcribed as "one one"

Is there anything I can do to improve the accuracy in such cases?

2. Is more of a question, is there a way to set up my grammar so it will only listen to 3 consecutive numbers, and will that improve the accuracy if the model is configured to listen to 3 consecutive numbers?

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.