Speech:Exps 0283 009

Description
Authors: Benjamin Leith James Schumacher Jon Shallow Ryan O'Neal

Date: 3/14/16

Purpose: We're running this experiment to determine if our new corpus will generate better WER results. Our new corpus (first_4hr) was generated with the createUtts.pl script which properly generates audio utterance files given a transcript file (from first_5hr corpus).

Details:
 * Started train at 2:59 pm on 3/14/16
 * Train finished 14 minutes later
 * Started decode at 3:20 PM on 3/14/16
 * Decode finished 15 minutes later

Results: SYSTEM SUMMARY PERCENTAGES by SPEAKER

,-.     |                            hyp.trans                            | |-|     | SPKR    | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err | |-+-+-|     | sw2001b |   18    163 | 79.8   17.2    3.1   44.2   64.4  100.0 | |-+-+-|     | sw2001a |   14    101 | 82.2   15.8    2.0   53.5   71.3  100.0 | |-+-+-|     | sw2005a |   39    701 | 83.6   10.7    5.7   13.3   29.7   94.9 | |-+-+-|     | sw2005b |   67    613 | 62.8   26.1   11.1   29.7   66.9  100.0 | |-+-+-|     | sw2006b |   29    618 | 74.6   17.6    7.8    7.3   32.7  100.0 | |-+-+-|     | sw2006a |   33    455 | 82.9   14.3    2.9   16.7   33.8  100.0 | |-+-+-|     | sw2007a |   61    614 | 78.7   13.5    7.8   16.1   37.5   91.8 | |-+-+-|     | sw2007b |   60    861 | 82.1   13.9    3.9    8.0   25.9   83.3 | |-+-+-|     | sw2008a |   24    260 | 75.4   22.3    2.3   25.8   50.4  100.0 | |-+-+-|     | sw2008b |   26    257 | 87.5   10.9    1.6   28.0   40.5   92.3 | |-+-+-|     | sw2009b |   23    181 | 78.5   14.9    6.6   30.4   51.9   95.7 | |-+-+-|     | sw2009a |   34    473 | 66.0   27.3    6.8   19.9   53.9  100.0 | |-+-+-|     | sw2010b |   22    404 | 69.3   22.8    7.9   12.1   42.8  100.0 | |-+-+-|     | sw2010a |   27    284 | 81.3   12.3    6.3   23.9   42.6   92.6 | |-+-+-|     | sw2012a |   45    838 | 81.1   12.6    6.2   13.5   32.3   97.8 | |-+-+-|     | sw2012b |   28    464 | 76.9   17.5    5.6   16.2   39.2  100.0 | |-+-+-|     | sw2013a |   35    377 | 67.9   26.0    6.1   21.8   53.8  100.0 | |-+-+-|     | sw2013b |   69    942 | 53.9   31.6   14.4   12.6   58.7   97.1 | |-+-+-|     | sw2014a |    9     79 | 75.9   19.0    5.1   32.9   57.0  100.0 | |-+-+-|     | sw2014b |   13    174 | 80.5   14.4    5.2   16.7   36.2  100.0 | |-+-+-|     | sw2015a |   21    375 | 74.7   15.7    9.6    3.7   29.1   90.5 | |-+-+-|     | sw2015b |   32    542 | 74.4   15.1   10.5    9.0   34.7   96.9 | |-+-+-|     | sw2017b |   38    658 | 81.2   13.5    5.3   14.4   33.3   97.4 | |-+-+-|     | sw2017a |   40    332 | 76.2   19.3    4.5   31.3   55.1  100.0 | |-+-+-|     | sw2018a |   49    505 | 57.6   35.8    6.5   41.4   83.8  100.0 | |-+-+-|     | sw2018b |   42    586 | 72.4   20.8    6.8   19.3   46.9  100.0 | |-+-+-|     | sw2019a |   50    594 | 76.6   16.5    6.9   15.5   38.9   98.0 | |-+-+-|     | sw2019b |   52    452 | 84.7   11.5    3.8   26.3   41.6  100.0 | |=================================================================|     | Sum/Avg | 1000  12903 | 74.6   18.6    6.9   18.1   43.5   97.0 | |=================================================================|     |  Mean   | 35.7  460.8 | 75.7   18.2    6.2   21.6   45.9   97.4 | | S.D.   | 16.3  229.0 |  8.0    6.4    2.9   11.8   14.2    4.1 | | Median | 33.5  459.5 | 76.8   16.2    6.2   18.0   42.1  100.0 | `-'