Speech:Exps 0227

From Openitware
Jump to: navigation, search

Description

Author:

Colby Johnson

Date:

19March2014

Purpose:

Test the mini/train accuracy against the 0192 (100hr/train) acoustic model.

Details:

The Language model was build around the full/train transcription. The mini/train is about less than an hour of data.

Results:

xRT = 3.81

                    SYSTEM SUMMARY PERCENTAGES by SPEAKER
     ,-----------------------------------------------------------------.
     |                            hyp.trans                            |
     |-----------------------------------------------------------------|
     | SPKR    | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
     |=================================================================|
     | Sum/Avg |  549  10774 | 78.0   17.4    4.6   12.6   34.6   96.0 |
     |=================================================================|
     |  Mean   |  2.9   57.0 | 78.1   18.2    3.7   21.5   43.4   97.1 |
     |  S.D.   |  1.9   44.5 | 12.9   11.7    4.1   29.8   31.4   10.8 |
     | Median  |  3.0   47.0 | 79.1   16.7    2.8   12.6   36.6  100.0 |
     `-----------------------------------------------------------------'

Successful Completion