:Units Paper

From SpeechWiki

Revision as of 18:01, 17 June 2009 (view source) Arthur (Talk | contribs) m ← Older edit		Revision as of 02:09, 21 June 2009 (view source) Arthur (Talk | contribs) Newer edit →
Line 79:		Line 79:
	1604		1604
	WER 49.8		WER 49.8
		+
		+	$ lg model/convGaus.wordInternal/trainTriUnit.splitConverge/triphone/train1/learnedParams.gmtk
		+	Finished reading in all parameters and structures
		+	Total number of trainable parameters in input files = 9856204
		+
		+	% dense PMFs
		+	1963
		+	WER 50.0
	==Debuging==		==Debuging==

---

Revision as of 02:09, 21 June 2009

Contents 1 Outline 2 Tests for units paper 2.1 put in above table 3 Debuging 4 Monophone tests

Outline

1. Intro
2. Unit Selection

   • Mistake instance

     Unit

     Replacement

   • Multwords
3. Baseline Description
   • Vocab: single most frequent pronunciation from a multi-pronunciation dictionary (better than multi-pronunciation)
4. Results
   • what to emphasize? Ideally, units+DTs will beat just DTs for every number of components. Even if we cannot grow the components until improvement bottoms out, at least there will be a trend.
5. Conclusion
   • Future work: consider context during unit selection (right now the unit is context-free - the same unit appearing in all contexts where replacements took place).

Tests for units paper

tests to run

compPer:	units:	monophone states	Mix:	totalComp:	WER	Test WER	Important
512	1		503	256k	TR
256	1		1000	256k	TR
64	1		3854	256k	49.3
64	2		2000	256k	?
32	4		2000	256k	?
32	2		4000	256k	?
alternatively
256	48	137	503	127971	53.0
128	48	137	1033	131185	50.9
32	48	137	3845	122907	51.4
64	112	615	2024	~128k	TR
16	4		2000	128k	?
16	112	615	4000	128k	?
my best clean baseline so far:
64	48	137	1033	65900	50.3%
64	48	137	1037	65900	50.6%		Finally, multi-unit with no units gives the same WER

put in above table

$ lg model/convGaus.noUnits/trainTriUnit.splitConverge/triphone/train1/learnedParams.gmtk Finished reading in all parameters and structures Total number of trainable parameters in input files = 5342017

% dense PMFs 1061 WER 51.6

$ lg model/convGaus.noTieWordInternal/trainTriUnit.splitConverge/triphone/train1/learnedParams.gmtk Finished reading in all parameters and structures Total number of trainable parameters in input files = 8084629

% dense PMFs 1604 WER 49.8

$ lg model/convGaus.wordInternal/trainTriUnit.splitConverge/triphone/train1/learnedParams.gmtk Finished reading in all parameters and structures Total number of trainable parameters in input files = 9856204

% dense PMFs 1963 WER 50.0

Debuging

A number of bugs have been fixed, and results on the SST:Units Paper Debugging are hopefully no longer relevant.

Monophone tests

Finally, at least something reasonable: and (hopefully) a 2% WER improvement in the baseline So redoing monophone tests:

monophone tests

trainConfig	testConfig	Descr	States	WER	WER 2k	Comments
triUnitsModel/config28	testc67triUnitsModelc28/config67	baseline but using the new testing stuff: --maxStates 0 --unitType wordInternalOnly --subUnits asBefore --growUnitSet 0	137	84.3%	86.0%
moreDTsModel/config20	trTestTrigramFixed/config19	baseline	137	84.3%		Same as above but using older code
triUnitsModel/config27	testc67TriOnBreakdownModelc27/config67	--maxStates 500 --unitType wordInternalOnly --subUnits asBefore --growUnitSet 0	627	87.2	87.6
triUnitsModel/config26	testc67TriOnBreakdownModelc26/config67	--maxStates 500 --unitType wordInternalOnly --subUnits asBefore	627	87.1 was 91.1%	TE
triUnitsModel/config25	xx	--maxStates 500 --unitType wordInternalOnly	636	xx	87.1
triUnitsModel/config24	xx	--maxStates 500	615	xx	xx
triUnitsModel/config29	xx	--maxStates 500 --unitType multiWordOnly	637	xx	88.2	There must be something wrong with this - the confusion pairs are not making sense and there are way too many deletions
triUnitsModel/config30	xx	--maxStates 500 --unitType wordInternalOnly --subUnits asBefore, initializeUnitsFromFile	627	xx	86.0
triUnitsModel/config31	xx	--maxStates 500 --contextType wordInternalOnly --subUnits asBefore --tieContext none	633	xx	85.8	finally a slight improvement.