Re: sphinx2-continuous, LMs, status

[robert b <http://dummy.us.eu.org/robert>]

--- Jonathan Young <http://www.attbi.com/~Jonathan_Young> wrote:
> Robert -
> 
> I still have not found any FE code (for any platform) which reads from a
> file except for the sphinx2-allphone code, which is broken for other reasons
> (requires context-independent phoneme models).  Can you clarify (again)
> how -adcin TRUE works?  I feel that this is probably an unproductive avenue
> of attack...

I'm attaching sphinx2-turtle (not an accurate name, sorry!).  It uses -adcin TRUE.

You'll have to modify the values in this script to point to the right files.

For me, this reads all the .wav files in the $DATA directory and feeds them into the recognizer.

Could you hack it up and try it out?  I just want to know if this tried to open the microphone
device.  If it does, I'll have to hack on sphinx2-continuous so it doesn't do that.

Thanks.

#!/bin/sh
S2BATCH=/usr/local/bin/sphinx2-continuous
#HMM=/usr/local/src/hub4-2002-11-05/s2-models
#HMM=/usr/local/src/hub4-s2-am/00.s2models
#HMM=/usr/local/src/hub4-2000-11-17-1/sphinx_2_format
HMM=model_parameters/time.s2models
#CTL=/usr/local/share/sphinx2/model/lm/turtle/turtle.ctl
#CTL=/usr/local/src/hub4-2002-11-05/sample.ctl
CTL=flatland/flat.fileids
#TASK=/usr/local/share/sphinx2/model/lm/full
#TASK=/usr/local/share/sphinx2/model/lm/turtle
TASK=/usr/local/src/hub4-2002-11-05/sample
#LM=/usr/local/share/sphinx2/model/lm/full/full.lm
#LM=/usr/local/share/sphinx2/model/lm/turtle/turtle.lm
LM=/usr/local/src/hub4-2002-11-05/sample/sample.7404.arpabo
#DICT=/usr/local/src/hub4-s2-am/train.dict
#DICT=/usr/local/share/sphinx2/model/lm/full/full.dic
#DICT=/usr/local/share/sphinx2/model/lm/turtle/turtle.dic
#DICT=etc/time.dic
DICT=/usr/local/src/hub4-2002-11-05/sample/sample.7404.sphinx
DATA=flatland/raw

echo " "
echo "sphinx2-turtle"
echo "Run CMU Sphinx2 in Batch mode to decode an example utterance."
echo " "

$S2BATCH -verbose 9 -adcin TRUE -adcext wav \
	-ctlfn ${CTL} -ctloffset 0 -ctlcount 100000000 \
	-datadir ${DATA} -agcmax TRUE -langwt 6.5 -fwdflatlw 8.5 \
	-rescorelw 9.5 -ugwt 0.5 -fillpen 1e-10 -silpen 0.005 -inspen 0.65 \
	-top 1 -topsenfrm 3 -topsenthresh -70000 -beam 2e-06 -npbeam 2e-06 \
	-lpbeam 2e-05 -lponlybeam 0.0005 -nwbeam 0.0005 -fwdflat FALSE \
	-fwdflatbeam 1e-08 -fwdflatnwbeam 0.0003 -bestpath TRUE \
	-kbdumpdir ${TASK} -lmfn ${LM} -dictfn ${DICT} \
	-noisedict ${HMM}/noisedict -phnfn ${HMM}/phone \
	-mapfn ${HMM}/map -hmmdir ${HMM} -hmmdirlist ${HMM} -8bsen TRUE \
	-sendumpfn ${HMM}/sendump -cbdir ${HMM}