Speech:Exps 0283 015

Description
Authors: James Schumacher

Date: 3/26/16

Purpose: Test the new utt files now in /mnt/main/corpus/switchboard/corpus/full/train/audio/utt/. [jrs1036@caesar utt]$ sox --i sw2001A-ms98-a-0002.sph

Input File    : 'sw2001A-ms98-a-0002.sph' Channels      : 1 Sample Rate   : 8000 Precision     : 32-bit Duration      : 00:00:10.58 = 84670 samples ~ 793.781 CDDA sectors File Size     : 340k Bit Rate      : 257k Sample Encoding: 32-bit Signed Integer PCM

[jrs1036@caesar utt]$ more sw2001A-ms98-a-0002.sph NIST_1A 1024  sample_count -i 84670 sample_n_bytes -i 4 channel_count -i 1 sample_byte_format -s2 01 sample_rate -i 8000 sample_coding -s3 pcm end_head Details:
 * Train configuration
 * Corpus: /mnt/main/corpus/switchboard/145hr
 * $CFG_VARNORM = 'no' (variance normalization)
 * $CFG_FINAL_NUM_DENSITIES = 8 (density)
 * $CFG_N_TIED_STATES = 1000 (senones)
 * $CFG_CONVERGENCE_RATIO = 0.04 (convergence ratio)
 * prepareTrainExperiment.pl took noticeably longer than the previous experiment. Unfortunately, I didn't time it because I was expecting it to be quick. However, I'd wager that it took about 10 min vs the 10 or so seconds it took last time.
 * Started generateFeats.pl at 5:50 PM on 3/26/2016
 * generateFeats.pl ended at 6:11 PM on 3/26/2016
 * generateFeats.pl also took quite a bit longer than the previous experiment. Not sure by how much, but it's definitely significant. This trend of the processes taking longer could be an indication that our results will be more accurate.
 * Started train at 6:16 PM on 3/26/2016
 * Train ended at 1:05 AM on 3/27/2016
 * Decode configuration
 * Decoding on: /mnt/main/corpus/switchboard/145hr/test/trans/train.trans
 * Decoding at: 1000 senones
 * Decode started at 1:53 AM on 3/27/2016
 * Decode ended at 3:31 AM on 3/27/2016

Results: SYSTEM SUMMARY PERCENTAGES by SPEAKER

,---.   |                             hyp.trans                             | |---|   | SPKR    | # Snt # Wrd | Corr    Sub    Del    Ins     Err   S.Err | |-+-+---|   |===================================================================|    | Sum/Avg | 3898  57265 | 61.4   29.2    9.4    8.5    47.1    91.7 | |===================================================================|   |  Mean   |  1.3   18.6 | 66.7   26.3    7.0   17.5    50.8    91.8 | | S.D.   |  0.5   15.9 | 21.0   18.1    9.1   47.5    49.0    25.6 | | Median |  1.0   14.0 | 66.7   26.7    3.3    5.2    45.2   100.0 | `---'
 * This is quite an improvement over the last experiment. This is 14.5% better than Exp 0283/014.