Timeshrinking
From SpeechWiki
(Difference between revisions)
Line 2: | Line 2: | ||
{| class="wikitable" | {| class="wikitable" | ||
- | ! <math>\tau</math> !! dev 2k utt WER | + | ! <math>\tau</math> !! dev 2k utt WER !! frames dropped |
|- | |- | ||
- | | .6 || 69.3% | + | | .6 || 69.3% || ~35% |
+ | |- | ||
+ | | .9 || xx || ~5% | ||
|} | |} | ||
Revision as of 06:27, 21 August 2009
Fisher experiments
<math>\tau</math> | dev 2k utt WER | frames dropped |
---|---|---|
.6 | 69.3% | ~35% |
.9 | xx | ~5% |
I have to check for bugs. It could be that the threshold is too low or it could be something else too.
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.