Timeshrinking
From SpeechWiki
(Difference between revisions)
Line 14: | Line 14: | ||
{| class="wikitable" | {| class="wikitable" | ||
|+ Timeshrinking results on fisher | |+ Timeshrinking results on fisher | ||
- | ! train <math>\tau</math> !! test <math>\tau</math> !! dev | + | ! train <math>\tau</math> !! test <math>\tau</math> !! dev 2000 utt WER !! dev 2000 utt on triphone single-gaussian model WER !! comments |
|- | |- | ||
- | | 1 || 1 || [{{ | + | | 1 || 1 || [{{FisherPath}}/exp/triphone/test2kUttOnConvGaus.noUnits/config73/LATEST.log 51.6%] || || old baseline |
|- | |- | ||
- | | 1 || 1 || [{{ | + | | 1 || 1 || [{{FisherPath}}/exp/.../LATEST.log TR] || [{{FisherPath}}/exp/timeshrink/test/unit.tri.timeshrink.1.onSingleGaussian/LATEST.log TE] || baseline rerun exactly as timeshrinking to really make sure it's not getting an unfair advantage |
|- | |- | ||
| .6 || .6 || 69.3% | | .6 || .6 || 69.3% | ||
|- | |- | ||
- | | .9 || .9 || [{{ | + | | .9 || .9 || [{{FisherPath}}/exp/timeshrink/test/unit.tri.timeshrink.point9/LATEST.log 56.3%] || [{{FisherPath}}/exp/timeshrink/test/triphoneSingleGausian/unit.tri.timeshrink.point9/LATEST.log 80.4%] |
|- | |- | ||
- | | 1 || .9 || [{{ | + | | 1 || .9 || [{{FisherPath}}/exp/timeshrink/testOnBaseline/unit.tri.timeshrink.point9/LATEST.log 53.9%] |
|- | |- | ||
- | | .9 || .9 || [{{ | + | | .9 || .9 || [{{FisherPath}}/exp/timeshrink/test/unit.tri.timeshrink.point9.noTsStr/LATEST.log 57.2] || [{{FisherPath}}/exp/timeshrink/test/triphoneSingleGausian/LATEST.log 80.7%] || using the non-timeshrinking str file for test |
+ | |- | ||
+ | | .9 || 1 || [{{FisherPath}}/exp/timeshrink/test/unit.tri.timeshrink.1.noTsStr/LATEST.log 54.6] | ||
+ | |- | ||
+ | | 1 || 1 || [{{FisherPath}}/exp/timeshrink/test/.../LATEST.log TR] || || PLP+MLP tandem | ||
|- | |- | ||
- | |||
|} | |} | ||
Revision as of 17:39, 24 September 2009
Fisher experiments
<math>\tau</math> | frames dropped |
---|---|
1 | 0% |
.9 | ~5% |
.6 | ~35% |
train <math>\tau</math> | test <math>\tau</math> | dev 2000 utt WER | dev 2000 utt on triphone single-gaussian model WER | comments |
---|---|---|---|---|
1 | 1 | 51.6% | old baseline | |
1 | 1 | TR | TE | baseline rerun exactly as timeshrinking to really make sure it's not getting an unfair advantage |
.6 | .6 | 69.3% | ||
.9 | .9 | 56.3% | 80.4% | |
1 | .9 | 53.9% | ||
.9 | .9 | 57.2 | 80.7% | using the non-timeshrinking str file for test |
.9 | 1 | 54.6 | ||
1 | 1 | TR | PLP+MLP tandem |
I have to check for bugs. It could be that the threshold is too low or it could be something else too. We should probably rerun baseline too, just to make sure I didn't optimize it unfairly.
Things to try
- Test svitchboard with fisher-trained model to see if we still get good results
- Train and test on plp+mlp, like svitchboard timeshrinking was done.
- Do baseline train+test to see if something changed in going from baseline to timeshrink structure files.
Future Directions
- Can be viewed as a two-mode special case of best-first viterbi search. So make a real best-first lattice search. Mark mentioned some attempts in the 80'ies to do this.