Ticket #85 (new defect)
HTK Adaptation of Acoustic Models not giving good results
Reported by: | kmaclean | Owned by: | kmaclean |
---|---|---|---|
Priority: | major | Milestone: | Acoustic Model 1.0 |
Component: | Acoustic Model | Version: | 0.1-alpha |
Keywords: | Cc: |
Description
Need to adjust HTK commands for adaptation to give better results.
Need to try with HTK 3.4 beta (still some problems with the alpha version). Never could get adaptation to work with HTK 3.3 see ticket #41.
Change History
Note: See
TracTickets for help on using
tickets.
may be easiest to create tutorial showing users how to combine their own audio files (converted to mfc) with the VoxForge? MFCs (which are smaller than wav files - therefore less to download) to create an acoustic model from scratch with the combined data - see ticket #78