Speech:Exps 0246

From Openitware
Jump to: navigation, search

Description

Author: Colby Johnson

Date: 3/26/14

Purpose: using what we know now, we will run a train and decode it to get the best percentage we can at this point in time.

Details: Decode on last_5hr/train acoustic model using the mini/train data set. LM was build using last_5hr/train data set.


Results: Decode: xRT = 1.84

                    SYSTEM SUMMARY PERCENTAGES by SPEAKER
     ,-----------------------------------------------------------------.
     |                            hyp.trans                            |
     |-----------------------------------------------------------------|
     | SPKR    | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
     |=================================================================|
     | Sum/Avg |  549  10774 | 40.6   52.0    7.4   15.7   75.1   99.3 |
     |=================================================================|
     |  Mean   |  2.9   57.0 | 42.5   51.0    6.6   22.3   79.8   99.0 |
     |  S.D.   |  1.9   44.5 | 14.4   13.5    7.0   29.7   28.6    8.2 |
     | Median  |  3.0   47.0 | 41.1   50.9    5.4   15.4   75.0  100.0 |
     `-----------------------------------------------------------------'

Successful Completion