[hts-users:00302] Re: training error - no proto model

[donall.mccann5@xxxxxxxxxxx]

Hi.
 I checked the files and full.list doesn't seem to include any -oy+ models,
but there is one mentioned in full_all.list. I removed it to see what would
happen and it has now given me a similar error, as follows:

// make unseen model
AU /home/dmccan-cl4/Desktop/Synthesis/mehts/data/lists/full_all.list
 Creating HMMset using trees to add unseen triphones
  ERROR [+2662]  FindProtoModel: no proto for l^eh-zh+er=ah@4_1/A:0_0_2/B:1-1-4@1-2&4-9#1-5$1-4!3-2;3-3|eh/C:0+0+1/D:det_1/E:content+2@3+6&2+3#2+2/F:in_1/G:3_2/H:12=8@6=6|L-L%/I:10=8/J:73+56-11
in hSet
 FATAL ERROR - Terminating program /usr/local/HTK-3.3/bin/HHEd
Error in /usr/local/HTK-3.3/bin/HHEd   -A -B -C /home/dmccan-cl4/Desktop/Synthesis/mehts/configs/trn.cnf
-D -T 1 -p -i -H /home/dmccan-cl4/Desktop/Synthesis/mehts/models/qst001/ver1/cmp/re_clustered.mmf
-w /home/dmccan-cl4/Desktop/Synthesis/mehts/models/qst001/ver1/cmp/re_clustered_all.mmf
/home/dmccan-cl4/Desktop/Synthesis/mehts/edfiles/qst001/ver1/cmp/mku.hed
/home/dmccan-cl4/Desktop/Synthesis/mehts/data/lists/full.list

If the program is searching for a model with the same center phoneme, could
it be a problem with the amount of training data (only 30 sentences) or am
I thinking along the wrong lines entirely?

Donall

>-- Original Message --
>Date: Wed, 17 May 2006 15:27:01 +0900
>From: "Heiga ZEN (Byung Ha CHUN)" <zen@xxxxxxxxxxxxxxxx>
>Reply-To: hts-users@xxxxxxxxxxxxxxxxxxxxxxxxx
>Subject: [hts-users:00301] Re: training error - no proto model
>To: hts-users@xxxxxxxxxxxxxxxxxxxxxxxxx
>
>
>Hi everyone,
>
># Now I'm in Toulouse to attend ICASSP2006 :-)
># Tomorrow I have a poster presentation about trajectory modeling for speech
>recognition.
>
>donall.mccann5@xxxxxxxxxxx wrote:
>> Hi. When I run the training script, I get the following error message:
>> 
>> // make unseen model
>> AU /home/dmccan-cl4/Desktop/Synthesis/mehts/data/lists/full_all.list
>>  Creating HMMset using trees to add unseen triphones
>>   ERROR [+2662]  FindProtoModel: no proto for ax^p-oy+n=t@1_3/A:0_0_2/B:1-0-3@3-2&7-2#3-1$3-2!2-0;2-1|oy/C:0+1+4/D:content_1/E:content+4@5+1&2+0#1+0/F:content_1/G:9_5/H:8=5@5=7|L-L%/I:4=3/J:71+55-11
>> in hSet
>>  FATAL ERROR - Terminating program /usr/local/HTK-3.3/bin/HHEd
>> Error in /usr/local/HTK-3.3/bin/HHEd   -A -B -C /home/dmccan-cl4/Desktop/Synthesis/mehts/configs/trn.cnf
>> -D -T 1 -p -i -H /home/dmccan-cl4/Desktop/Synthesis/mehts/models/qst001/ver1/cmp/re_clustered.mmf
>> -w /home/dmccan-cl4/Desktop/Synthesis/mehts/models/qst001/ver1/cmp/re_clustered_all.mmf
>> /home/dmccan-cl4/Desktop/Synthesis/mehts/edfiles/qst001/ver1/cmp/mku.hed
>> /home/dmccan-cl4/Desktop/Synthesis/mehts/data/lists/full.list
>> 
>> Where should I include the required proto model and how should I go about
>> making it? The only proto model I know of is the one in the proto directory...
>> (sorry if this is something obvious! I get a bit lost whenever I get too
>> deep into HTS).
>
>This error seems to indicate that there is no *-oy+* models in the model
>set.
>To generate an unseen model, HTK searches a model whose center phoneme is
>equal to that of the unseen model to specify 
>its model topology (#states, etc.) and associate its transition probability
>matrix.
>Please check your full.list (not full_all.list) whether it includes any
*-oy+*
>models.
>If your full.list contains any *-oy+* models, this may be a bug of the HTS
>implementation.
>
>Regards,
>
>Heiga Zen (Byung Ha Chun)
>
>-- 
>------------------------------------------------
>  Heiga ZEN     (in Japanese pronunciation)
>  Byung Ha CHUN (in Korean pronunciation)
>
>  Department of Computer Science and Engineering
>  Nagoya Institute of Technology
>  Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan
>
>  http://kt-lab.ics.nitech.ac.jp/~zen
>------------------------------------------------
>