VoxForge
Hi all, I am creating a new acoustic/language model to recognise alphanumeric utterances from a speech synthesiser. Preliminary training completes, but at the decode stage i get this error from slave.pl: MODULE: DECODE Decoding using models previously trained Aligning results to find error rate Can't open ./results/festvox-1-1.match word_align.pl failed with error code 65280 at ./scripts_pl/decode/slave.pl line 172. Digging into festvox.align i get this error: UttID is ignored but file size mismatch between /home/cartel/SphinxTrain/festvox/etc/festvox_test.transcription11978 and ./results/festvox.match11978 at ./scripts_pl/decode/word_align.pl line 55. Does anyone have any ideas? Something in decode.cfg? Thanks in advance
--- (Edited on 2/28/2009 10:30 pm [GMT-0600] by cartel) ---
So the issue is resolved and the problem was in not running the setup
sphinx3/scripts/setup_sphinx3.pl
--- (Edited on 3/1/2009 3:48 am [GMT-0600] by nsh) ---
Hey nsh
I tried that but it still doesnt work. Any other suggestions.
Cheers
Hozefa
--- (Edited on 6/8/2009 10:06 am [GMT-0500] by hozefa26) ---
I have the same problem:
>perl scripts_pl/decode/slave.pl
MODULE: DECODE Decoding using models previously trained
Decoding 130 segments starting at 0 (part 1 of 1)
0%
Aligning results to find error rate
Can't open /home/xpecttrum/sphinx_tutorial/an4/result/an4-1-1.match
word_align.pl failed with error code 65280 at scripts_pl/decode/slave.pl line 173.
-------------------------------
I could not find the reason yet, anyone?
--- (Edited on 10/9/2009 7:33 am [GMT-0500] by Visitor) ---
anyone tells you to look in the log in logdir/decode and read it carefully.
--- (Edited on 10/9/2009 16:56 [GMT+0400] by nsh) ---
Thanks snh,
This helps, now I realised also that I can check some messages at my local file
file:///home/xpecttrum/sphinx_tutorial/an4/an4.html
looking to the logfile I could solve 2 issues I had, now the program runs.
However, I still don't have the desired results, it says me that this step had 6 errors and 2 warnings...
---
Question: Is it going to decode actually what sound?
--- (Edited on 10/9/2009 8:31 am [GMT-0500] by Visitor) ---
I will post here because maybe helps someone else.
In my case, the solution for this error was:
editing /etc/ld.so.conf
add a new line with "/usr/local/lib" or the path directory that contains the file "libsphinxbase.so.1"
then run "sudo ldconfig"
That's it.
--- (Edited on 10/9/2009 9:24 am [GMT-0500] by Visitor) ---
I will post here because maybe helps someone else.
In my case, the solution for this error was:
editing /etc/ld.so.conf
add a new line with "/usr/local/lib" or the path directory that contains the file "libsphinxbase.so.1"
then run "sudo ldconfig"
That's it.
--- (Edited on 10/9/2009 9:24 am [GMT-0500] by Visitor) ---
I forgot to post that I also needed to do this:
cp /home/xpecttrum/sphinx_tutorial/an4/etc/an4.ug.lm.DMP /home/xpecttrum/sphinx_tutorial/an4/etc/an4.lm.DMP
Don't know if it is here that I should post it, but a nice hint goes,
for listenning the .sph sounds, I am using:
aplay an4/wav/an4test_clstk/fcaw/*.sph -f S16_LE -c1 -r18000
I hope it is usefull to someone
--- (Edited on 10/9/2009 10:04 am [GMT-0500] by Visitor) ---