fast speaking

Speech Recognition Toolkit

Brought to you by: air, arthchan2003, awb, bhiksha, and 5 others

This project can now be found here.

fast speaking

Forum: Help

Creator: ignis10

Created: 2017-04-10

Updated: 2017-04-11

ignis10 - 2017-04-10

Hi,

I have a recording of a person speaking slow and the accuracy is fine,
when he turn to speak fast (i.e. 2-3 wps) the accuracy decreases dramatically,
Is there any configuration for this case?
any pre-processing?
anything else that could be done?

Thanks

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Nickolay V. Shmyrev - 2017-04-10
  
  There is no configuration or preprocessing, you have to adapt an acoustic model or train a new one.
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

ignis10 - 2017-04-11

thanks Nickolay,

if that so, and I have just 10 minutes of this fast speaking person,
would you recommand adapting (map_adapt) or use the mllr transfrom ?

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Nickolay V. Shmyrev - 2017-04-11
  
  Map should be fine.
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Log in to post a comment.