I train my data in quiet room, and the accuracy have 90%.
But when I use this model to test in noise environment(such as restaurant, outdoor, ...),
the recognition rate only 40%~50%.
Does there have solution that can solve this problem, if I don't want add noise data in my quiet model ?
Thanks for your help.
Last edit: stevenyslin 2016-07-01
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I train my data in quiet room, and the accuracy have 90%.
But when I use this model to test in noise environment(such as restaurant, outdoor, ...),
the recognition rate only 40%~50%.
Does there have solution that can solve this problem, if I don't want add noise data in my quiet model ?
Thanks for your help.
Last edit: stevenyslin 2016-07-01
There are big books on methods on robust speech recognition, there is no short answer, you can start with:
a review
https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/double_column.pdf
more complete book
http://store.elsevier.com/Robust-Automatic-Speech-Recognition/Jinyu-Li/isbn-9780128026168/