Problem in adapting acoustic model

[Jack Smith]

Hi
When running bw I am getting:
ERROR: "baum_welch.c", line 324: arctic_0001 ignored for all 20 wav files.

Platform is Jessie on Raspberry pi 3
Downloaded sphinxbase-5prealpha , sphinxtrain-5prealpha , pocketsphinx-5prealpha
Downloaded Model named cmusphinx-en-us-ptm-5.2.tar.gz from https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/US%20English/cmusphinx-en-us-ptm-5.2.tar.gz/download
Generated acoustic feature files by following the tutorial
Converted mdef file in en-us by following tutorial.
Ran bw with parameters per the tutorial:

./bw \
  -hmmdir en-us \
  -moddeffn en-us/mdef.txt \
  -ts2cbfn .ptm. \
  -feat 1s_c_d_dd \
  -svspec 0-12/13-25/26-38 \
  -cmn current \
  -agc none \
  -dictfn cmudict-en-us.dict \
  -ctlfn arctic20.fileids \
  -lsnfn arctic20.transcription \
  -accumdir .

I have reviewed and solved problems from several posts that report this same issue, but still have the problem:
The wav files were recorded using Audacity and appear to have the correct format:

$ sox --i arctic_0001.wav
Input File     : 'arctic_0001.wav'
Channels       : 1
Sample Rate    : 16000
Precision      : 16-bit
Duration       : 00:00:04.08 = 65294 samples ~ 306.066 CDDA sectors
File Size      : 131k
Bit Rate       : 256k
Sample Encoding: 16-bit Signed Integer PCM

I have removed leading and trailing blank space in the wav files. Good quality on the recordings – no clicks
I am using the sample arctic20.fileids and arctic20.trancription files. The recordings match the text in arctic20.trancription.
I checked that the feat.params matches the bw caoomand line info.
The files are here: https://www.dropbox.com/s/3x1e0tb1x1kr58m/forpostingissue.zip?dl=0
To save space, did not include en-us-lm.bin
The bw command line data is in the file bw cmd line info.txt
Thanks!

[Nickolay V. Shmyrev]

Please build everything from git

Please also provide information about hardware architecture you have. Run uname -a. Do you run something like big endian by chance?

[Jack Smith]

Thanks, Nickolay. Here is HW and OS info - not running Big Endian

uname -a
Linux raspberrypi 4.9.28-v7+ #998 SMP Mon May 15 16:55:39 BST 2017 armv7l GNU/Linux

pi@raspberrypi:~/jackwork $ lscpu
Architecture: armv7l
Byte Order: Little Endian
CPU(s): 4
On-line CPU(s) list: 0-3
Thread(s) per core: 1
Core(s) per socket: 4
Socket(s): 1
Model name: ARMv7 Processor rev 4 (v7l)
CPU max MHz: 1200.0000
CPU min MHz: 600.0000

In the process of building Sphinxbase, pocketsphinx, and sphinxtrain from git.

[Jack Smith]

Nickolay,

I re-installed sphinxbase, pocketsphinx, and sphinxtrain through git, as you requested
cd ~/
git clone git://github.com/cmusphinx/sphinxbase.git
cd ./sphinxbase
./autogen.sh
./configure --enable-fixed
make clean all
make check
sudo make install
git clone git://github.com/cmusphinx/pocketsphinx.git
cd ./pocketsphinx
./autogen.sh
./configure
make clean all
make check
sudo make install
git clone git://github.com/cmusphinx/sphinxtrain.git
cd ./sphinxtrain
./autogen.sh
./configure
make clean all
make check
sudo make install

However, the problem is not solved. Console results still have the ERROR: "baum_welch.c", line 324: arctic_0001 ignored for all 20 wav files https://www.dropbox.com/s/x9c97yrik3a23io/console_results%20after%20reinstall.txt?dl=0
Thanks!

[G10DRAS]

Do not use below while compiling sphinxbase on Raspberry Pi.
./configure --enable-fixed
Simply use
./configure
and recompile sphinxbase, pocketsphinx and sphinxtrain.

[Jack Smith]

Thanks!! That solved the problem - now not getting that error with bw