testing simon http://spirit.blau.in/simon my first steps with the simon speech recognition software Thu, 02 Sep 2010 17:34:10 +0000 en hourly 1 http://wordpress.org/?v=3.0.1 git pull origin master http://spirit.blau.in/simon/2010/09/02/git-pull-origin-master-3/ http://spirit.blau.in/simon/2010/09/02/git-pull-origin-master-3/#comments Thu, 02 Sep 2010 17:33:30 +0000 producer http://spirit.blau.in/simon/?p=5053 FETCH_HEAD Updating cbc744d..4ae4550 Fast-forward CMakeLists.txt | 6 +- cmake/FindPortaudio.cmake | [...]]]> This happens in my Ubuntu terminal:

ubuntu@ubuntu-desktop:~$ cd Documents/201005/speech2text && git pull origin master
remote: Counting objects: 66, done.
remote: Compressing objects: 100% (41/41), done.
remote: Total 41 (delta 30), reused 0 (delta 0)
Unpacking objects: 100% (41/41), done.
From git://speech2text.git.sourceforge.net/gitroot/speech2text/speech2text
* branch master -> FETCH_HEAD
Updating cbc744d..4ae4550
Fast-forward
CMakeLists.txt | 6 +-
cmake/FindPortaudio.cmake | 1 +
package/KDE/CMakeLists.txt | 92 +-
package/NSIS/anpassungen.nsi | 57 +-
simon/doc/de/index.docbook | 4183 +++++++++++++-------
simon/doc/en/index.docbook | 2 +-
simon/doc/po/simonmanual.pot | 1510 ++++----
simon/doc/po/simonmanual_de.po | 1979 +++++-----
.../src/simonmodelmanagementui/ImportDict/dict.cpp | 2 +-
.../ImportTrainingData/importtrainingdata.cpp | 31 +-
.../ImportTrainingData/importtrainingdata.h | 2 +
version.h | 2 +-
12 files changed, 4580 insertions(+), 3287 deletions(-)
ubuntu@ubuntu-desktop:~/Documents/201005/speech2text$

Great:

“+ allowedFileTypes << "wav";

+ allowedFileTypes << "mp3";

+ allowedFileTypes << "ogg";

+ allowedFileTypes << "flac";"

I am planning to release ogg files (because I don’t have unlimited webspace, and because I think that ogg quality is OK for training).

]]> http://spirit.blau.in/simon/2010/09/02/git-pull-origin-master-3/feed/ 1 Ralf’s German dictionary 0.2.1 http://spirit.blau.in/simon/2010/08/30/ralfs-german-dictionary-0-2-1/ http://spirit.blau.in/simon/2010/08/30/ralfs-german-dictionary-0-2-1/#comments Mon, 30 Aug 2010 21:49:17 +0000 producer http://spirit.blau.in/simon/?p=4887 How I build Ralf's German dictionary version 0.2.1:

1. Add rule to improve-german.xsl:

<xsl:when test="contains(lower-case(../grapheme), 'münd')"><xsl:value-of select="replace($sierra, 'mynd','mʏnd')"/></xsl:when>

2. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'fünf')"><xsl:value-of select="replace($sierra, 'fynf','fʏnf')"/>

3. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'hül')"><xsl:value-of select="replace($sierra, 'hyl','hʏl')"/></xsl:when>

4. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'rück')"><xsl:value-of select="replace($sierra, 'ʀyk','ʀʏk')"/></xsl:when>

5. The German ‘r’ is more like a vowel in this context:

<xsl:when test="contains(lower-case(../grapheme), 'behörde')"><xsl:value-of select="replace($sierra, 'bəhœʀd','bəhøːɐ̯d')"/></xsl:when>

6. The v has to be replaced by f:

<xsl:when test="contains(lower-case(../grapheme), 'vertret')"><xsl:value-of select="replace($sierra, 'vɐtʀət','fɐtʀeːt')"/></xsl:when>

7. The i is pronounced long:

<xsl:when test="contains(lower-case(../grapheme), 'widrig')"><xsl:value-of select="replace($sierra, 'vɪdʀɪg','viːdʀɪg')"/></xsl:when>

8. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'nüchter')"><xsl:value-of select="replace($sierra, 'nyçt','nʏçt')"/></xsl:when>

9. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'plünder')"><xsl:value-of select="replace($sierra, 'plyndəʀ','plʏndəʀ')"/></xsl:when>

10. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'rück')"><xsl:value-of select="replace($sierra, 'ʀyk','ʀʏk')"/></xsl:when>

11. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'rüst')"><xsl:value-of select="replace($sierra, 'ʀyst','ʀʏst')"/></xsl:when>

12. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'bewerb')"><xsl:value-of select="replace($sierra, 'beːvəʀb','beːvɛʀb')"/></xsl:when>

13. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'schmück')"><xsl:value-of select="replace($sierra, 'ʃmyk','ʃmʏk')"/></xsl:when>

14. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'schüsse')"><xsl:value-of select="replace($sierra, 'ʃysə','ʃʏsə')"/></xsl:when>

15. Add rule:

<xsl:when test="contains(lower-case(../grapheme), 'schütt')"><xsl:value-of select="replace($sierra, 'ʃyt','ʃʏt')"/></xsl:when>

16. Convert 287 phoneme sequences:

<xsl:when test="contains(lower-case(../grapheme), 'türm')"><xsl:value-of select="replace($sierra, 'tyʀm','tʏʀm')"/></xsl:when>

17. Convert 183 phonemes:

<xsl:when test="ends-with(lower-case(../grapheme), 'berg')"><xsl:value-of select="replace($sierra, 'bəʀk','bəʀk')"/></xsl:when>

18: Generate the next version with Ubuntu terminal:

cat '/home/ubuntu/Documents/201008/german-0.2.1/german-0.2.xml' | saxonb-xslt -ext:on -s:- \
-xsl:'/home/ubuntu/Documents/201008/german-0.2.1/improve-german.xsl' \
-o:'/home/ubuntu/Documents/201008/german-0.2.1/object/german-0.2.1.xml'

19. Convert 142 phonemes:

<xsl:when test="contains(lower-case(../grapheme), 'struktur')"><xsl:value-of select="replace($sierra, 'stʀʊktuː','ʃtʀʊktuː')"/></xsl:when>

20. Convert the Ach-Laut:

<xsl:variable name="sierra" select='replace($sierra, "x", "χ")'/>

21. Download and import Ralf's German dictionary.

]]> http://spirit.blau.in/simon/2010/08/30/ralfs-german-dictionary-0-2-1/feed/ 0 SAMPA of Ach-Laut: [X] instead of [x] http://spirit.blau.in/simon/2010/08/30/sampa-of-ach-laut-x-instead-of-x/ http://spirit.blau.in/simon/2010/08/30/sampa-of-ach-laut-x-instead-of-x/#comments Mon, 30 Aug 2010 21:28:09 +0000 producer http://spirit.blau.in/simon/?p=5032 Here is the result after importing a dictionary that uses the IPA symbol [χ] instead of the ASCII symbol [x]:

ach-lautThe SAMPA symbol is upper-case [X], and not lower-case [x]. This is OK, I think.

The next version of Ralf's German dictionary will employ the IPA symbol [χ] instead of [x].

]]> http://spirit.blau.in/simon/2010/08/30/sampa-of-ach-laut-x-instead-of-x/feed/ 0 samview.cpp http://spirit.blau.in/simon/2010/08/30/samview-cpp/ http://spirit.blau.in/simon/2010/08/30/samview-cpp/#comments Mon, 30 Aug 2010 14:24:30 +0000 producer http://spirit.blau.in/simon/?p=5027 This is what I am doing in the Ubuntu terminal:

ubuntu@ubuntu-desktop:~$ cd Documents/201005/speech2text && git pull origin master
remote: Counting objects: 15, done.
remote: Compressing objects: 100% (8/8), done.
remote: Total 8 (delta 7), reused 0 (delta 0)
Unpacking objects: 100% (8/8), done.
From git://speech2text.git.sourceforge.net/gitroot/speech2text/speech2text
* branch master -> FETCH_HEAD
Updating 5853434..4d72b2a
Fast-forward
sam/src/samview.cpp | 2 +-
simonlib/simonmodeltest/fileresultmodel.cpp | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
ubuntu@ubuntu-desktop:~/Documents/201005/speech2text$ ./build_ubuntu.sh

I will see later whether the sam Edit sample button has been fixed.

]]> http://spirit.blau.in/simon/2010/08/30/samview-cpp/feed/ 2 ./build_ubuntu.sh http://spirit.blau.in/simon/2010/08/29/build_ubuntu-sh/ http://spirit.blau.in/simon/2010/08/29/build_ubuntu-sh/#comments Sun, 29 Aug 2010 21:02:07 +0000 producer http://spirit.blau.in/simon/?p=5017 A few seconds ago:

ubuntu@ubuntu-desktop:~/Documents/201005/speech2text$ git pull origin masterremote: Counting objects: 17, done.
remote: Compressing objects: 100% (9/9), done.
remote: Total 9 (delta 8), reused 0 (delta 0)
Unpacking objects: 100% (9/9), done.
From git://speech2text.git.sourceforge.net/gitroot/speech2text/speech2text
* branch master -> FETCH_HEAD
Updating afe2203..5853434
Fast-forward
simon/model/phonemes | 1 +
.../src/simonmodelmanagementui/ImportDict/dict.cpp | 3 ---
2 files changed, 1 insertions(+), 3 deletions(-)
ubuntu@ubuntu-desktop:~/Documents/201005/speech2text$ ./build_ubuntu.sh

Let’s take a look into sam:

sam-edit-sample1. Sorting is now working. Thanks for fixing that issue.
2. But the Edit sample button seems to be broken now.

I imported the words Hochkulturen (with IPA-[χ]) and Hochkultur (with ASCII-[x]), but only the word Hochkultur is being displayed:

hochkulturAfter importing the words Hochkulturen (with UTF-8-[χ]) and Hochkultur (with ASCII-[x]), only one word remains. Obviously, the other word Hochkulturen has been omitted by simon. Maybe this is a simon security precaution.

]]> http://spirit.blau.in/simon/2010/08/29/build_ubuntu-sh/feed/ 1 git pull origin master http://spirit.blau.in/simon/2010/08/29/git-pull-origin-master-2/ http://spirit.blau.in/simon/2010/08/29/git-pull-origin-master-2/#comments Sun, 29 Aug 2010 19:52:51 +0000 producer http://spirit.blau.in/simon/?p=4998 This is what I did a few minutes ago because sorting has been fixed and because of the symbol [χ]:

ubuntu@ubuntu-desktop:~$ cd Documents/201005/speech2text && git pull origin master
remote: Counting objects: 35, done.
remote: Compressing objects: 100% (18/18), done.
remote: Total 18 (delta 15), reused 0 (delta 0)
Unpacking objects: 100% (18/18), done.
From git://speech2text.git.sourceforge.net/gitroot/speech2text/speech2text
* branch master -> FETCH_HEAD
Updating 97be4d2..afe2203
Fast-forward
sam/src/samview.cpp | 1 +
.../src/simonmodelmanagementui/ImportDict/dict.cpp | 3 +++
simonlib/simonmodeltest/fileresultmodel.cpp | 7 +++++--
3 files changed, 9 insertions(+), 2 deletions(-)
ubuntu@ubuntu-desktop:~/Documents/201005/speech2text$ ./build_ubuntu.sh

I want to see whether the Ach-Laut is imported correctly by importing these entries (first entry with ASCII-[x]; second entry with IPA-[χ]:

<lexeme role="Substantiv">
<grapheme>Hochkultur</grapheme>
<phoneme>hoːxkʊltuːʀ</phoneme>
<phoneme>hoːxkʊltuːɐ̯</phoneme>
</lexeme>
<lexeme role="Substantiv">
<grapheme>Hochkulturen</grapheme>
<phoneme>hoːχkʊltuːʀən</phoneme>
<phoneme>hoːχkʊltuːɐ̯n</phoneme>
</lexeme>

No, it didn’t work out:

ach-laut The word Hochkultur is displayed with the correct SAMPA. But the word Hochkulturen doesn’t have the correct SAMPA.

]]> http://spirit.blau.in/simon/2010/08/29/git-pull-origin-master-2/feed/ 0 sam: sorting should be improved http://spirit.blau.in/simon/2010/08/29/sam-sorting-should-be-improved/ http://spirit.blau.in/simon/2010/08/29/sam-sorting-should-be-improved/#comments Sun, 29 Aug 2010 08:14:14 +0000 producer http://spirit.blau.in/simon/?p=4986 A few minutes ago, I tested section ‘xea’ against all. The overall recognition rate was about 95%.

recognition-rateTake a closer look at the right column: The word DIVERGIERST has a recognition rate of 100%. The word DIN has a recognition rate of 11.9569%. You can see that the sorting is not correct.

Btw, it would be interesting to export the recognition results into a text file. What about adding an Export test results button?

]]> http://spirit.blau.in/simon/2010/08/29/sam-sorting-should-be-improved/feed/ 2 sam: test section ‘xea’ against all http://spirit.blau.in/simon/2010/08/29/sam-test-section-xea-against-all/ http://spirit.blau.in/simon/2010/08/29/sam-test-section-xea-against-all/#comments Sun, 29 Aug 2010 07:16:35 +0000 producer http://spirit.blau.in/simon/?p=4973 At the moment, I am testing with sam the speech model xea (contains about 800 German words) against Ralf’s German speech model 0.1.6 (contains about 25000 German words). The file:///home/ubuntu/Documents/201008/sam-0.1.6/test-xea-against-all.sam has the following content:

/home/ubuntu/.kde/share/apps/simon/model/hmmdefs
/home/ubuntu/.kde/share/apps/simon/model/tiedlist
/home/ubuntu/.kde/share/apps/simon/model/model.dict
/home/ubuntu/.kde/share/apps/simon/model/model.dfa
/home/ubuntu/Documents/201008/wav-all
/home/ubuntu/Documents/201006/audacity/xea-folder/wav-xea
/home/ubuntu/Documents/201008/sam-0.1.6/lexicon
/home/ubuntu/Documents/201008/sam-0.1.6/model.grammar
/home/ubuntu/Documents/201008/sam-0.1.6/simple.voca
/home/ubuntu/Documents/201008/prompts-wav.txt
/home/ubuntu/Documents/201006/audacity/xea-folder/prompts-xea
/usr/share/kde4/apps/simon/model/tree1.hed
/usr/share/kde4/apps/simon/model/wav_config
16000
/usr/share/kde4/apps/simond/default.jconf
2

Question: what does the 2 in the last line mean?

By the way, I used the Serialize scenarios button (input file:///home/ubuntu/Documents/201007/german-speech-model-0.1.6/scenario-0.1.6.xml) to generate the following files:
file:///home/ubuntu/Documents/201008/sam-0.1.6/lexicon
file:///home/ubuntu/Documents/201008/sam-0.1.6/model.grammar
file:///home/ubuntu/Documents/201008/sam-0.1.6/simple.voca

I think that it is a good idea to test a subset (e.g. xea) against the superset (xaa+xab+xac+…+xpw).

]]> http://spirit.blau.in/simon/2010/08/29/sam-test-section-xea-against-all/feed/ 0 Ralf’s German speech model 0.1.6 http://spirit.blau.in/simon/2010/08/28/ralfs-german-speech-model-0-1-6/ http://spirit.blau.in/simon/2010/08/28/ralfs-german-speech-model-0-1-6/#comments Sat, 28 Aug 2010 16:16:59 +0000 producer http://spirit.blau.in/simon/?p=4954 How I create Ralf’s German speech model 0.1.6:

1. Concatenate the different prompts file into one single prompts file:

cat /home/ubuntu/Documents/201006/audacity/xaa-folder/prompts-xaa \
/home/ubuntu/Documents/201006/audacity/xab-folder/prompts-xab \
/home/ubuntu/Documents/201006/audacity/xac-folder/prompts-xac \
/home/ubuntu/Documents/201006/audacity/xad-folder/prompts-xad \
/home/ubuntu/Documents/201006/audacity/xba-folder/prompts-xba \
/home/ubuntu/Documents/201006/audacity/xca-folder/prompts-xca \
/home/ubuntu/Documents/201006/audacity/xda-folder/prompts-xda \
/home/ubuntu/Documents/201006/audacity/xdb-folder/prompts-xdb \
/home/ubuntu/Documents/201006/audacity/xea-folder/prompts-xea \
/home/ubuntu/Documents/201006/audacity/xeb-folder/prompts-xeb \
/home/ubuntu/Documents/201006/audacity/xfa-folder/prompts-xfa \
/home/ubuntu/Documents/201006/audacity/xfb-folder/prompts-xfb \
/home/ubuntu/Documents/201006/audacity/xga-folder/prompts-xga \
/home/ubuntu/Documents/201006/audacity/xgb-folder/prompts-xgb \
/home/ubuntu/Documents/201006/audacity/xgx-folder/prompts-xgx \
/home/ubuntu/Documents/201006/audacity/xha-folder/prompts-xha \
/home/ubuntu/Documents/201006/audacity/xia-folder/prompts-xia \
/home/ubuntu/Documents/201006/audacity/xja-folder/prompts-xja \
/home/ubuntu/Documents/201006/audacity/xka-folder/prompts-xka \
/home/ubuntu/Documents/201006/audacity/xla-folder/prompts-xla \
/home/ubuntu/Documents/201006/audacity/xma-folder/prompts-xma \
/home/ubuntu/Documents/201006/audacity/xna-folder/prompts-xna \
/home/ubuntu/Documents/201006/audacity/xoa-folder/prompts-xoa \
/home/ubuntu/Documents/201006/audacity/xpa-folder/prompts-xpa \
/home/ubuntu/Documents/201006/audacity/xpb-folder/prompts-xpb \
/home/ubuntu/Documents/201006/audacity/xhb-folder/prompts-xhb \
/home/ubuntu/Documents/201006/audacity/xib-folder/prompts-xib \
/home/ubuntu/Documents/201006/audacity/xjb-folder/prompts-xjb \
/home/ubuntu/Documents/201006/audacity/xpw-folder/prompts-xpw \
/home/ubuntu/Documents/201006/audacity/xkb-folder/prompts-xkb \
/home/ubuntu/Documents/201006/audacity/xlb-folder/prompts-xlb \
/home/ubuntu/Documents/201006/audacity/xbm-folder/prompts-xbm \
/home/ubuntu/Documents/201006/audacity/xmb-folder/prompts-xmb \
/home/ubuntu/Documents/201006/audacity/xaae-folder/prompts-xaae \
> /home/ubuntu/Documents/201008/german-0.2.1./prompts-all

2. Delete old simon configuration files:

rm -r /home/ubuntu/.kde/share/apps/simon
rm -r /home/ubuntu/.kde/share/apps/simond
rm /home/ubuntu/.kde/share/config/sscrc
rm /home/ubuntu/.kde/share/config/speechmodelmanagementrc
rm /home/ubuntu/.kde/share/config/simonsoundrc
rm /home/ubuntu/.kde/share/config/simonscenariosrc
rm /home/ubuntu/.kde/share/config/simonrecognitionrc
rm /home/ubuntu/.kde/share/config/simonrc
rm /home/ubuntu/.kde/share/config/simoncommandrc
rm /home/ubuntu/.kde/share/config/samrc
rm /home/ubuntu/.kde/share/config/ksimondrc

3. Ubuntu terminal:

dir /home/ubuntu/Documents/201008/wav-all > /home/ubuntu/Documents/201008/german-0.2.1/prompts-wav-dir

4. Ubuntu terminal:

ubuntu@ubuntu-desktop:~$ saxonb-xslt -ext:on -s:'/home/ubuntu/Documents/201008/german-0.2.1/german-0.2.1.xml' -xsl:'/home/ubuntu/Documents/201008/german-0.2.1/compare-s-0.2.1-s-dir.xsl' -o:'/home/ubuntu/Documents/201006/audacity/xaa-folder/dummy.xml'

5. Download Ralf's German speech model 0.1.6.

]]> http://spirit.blau.in/simon/2010/08/28/ralfs-german-speech-model-0-1-6/feed/ 0 Ralf’s German speech model 0.1.5 http://spirit.blau.in/simon/2010/08/26/ralfs-german-speech-model-0-1-5/ http://spirit.blau.in/simon/2010/08/26/ralfs-german-speech-model-0-1-5/#comments Thu, 26 Aug 2010 17:20:44 +0000 producer http://spirit.blau.in/simon/?p=4933 How I create Ralf's German speech model version 0.1.5 (download):

1. Delete old directories (cp-xmb):

rm -r /home/ubuntu/.kde/share/apps/simon
rm -r /home/ubuntu/.kde/share/apps/simond
rm /home/ubuntu/.kde/share/config/sscrc
rm /home/ubuntu/.kde/share/config/speechmodelmanagementrc
rm /home/ubuntu/.kde/share/config/simonsoundrc
rm /home/ubuntu/.kde/share/config/simonscenariosrc
rm /home/ubuntu/.kde/share/config/simonrecognitionrc
rm /home/ubuntu/.kde/share/config/simonrc
rm /home/ubuntu/.kde/share/config/simoncommandrc
rm /home/ubuntu/.kde/share/config/samrc
rm /home/ubuntu/.kde/share/config/ksimondrc

2. Ubuntu terminal:

ubuntu@ubuntu-desktop:~$ saxonb-xslt -ext:on -s:/home/ubuntu/Documents/201006/audacity/xmb-folder/lexicon-xmb.xml -xsl:/home/ubuntu/Documents/201006/audacity/copy-flac.xsl -o:/home/ubuntu/Documents/201006/audacity/xfa-folder/cp-dummy

This command produced the file:///home/ubuntu/Documents/201008/lexicon-all.xml with 24000 German words.

3. Concatenate all prompts in one single file (cp-xmb):

cat /home/ubuntu/Documents/201006/audacity/xaa-folder/prompts-xaa \
/home/ubuntu/Documents/201006/audacity/xab-folder/prompts-xab \
/home/ubuntu/Documents/201006/audacity/xac-folder/prompts-xac \
[...]
/home/ubuntu/Documents/201006/audacity/xbm-folder/prompts-xbm \
/home/ubuntu/Documents/201006/audacity/xmb-folder/prompts-xmb \
> /home/ubuntu/Documents/201008/prompts-all

4. Convert FLAC files into wav format (cp-xmb):

cd /home/ubuntu/Documents/201006/audacity/xaa-folder/flac-xaa && \
for f in *.flac; do sox "$f" /home/ubuntu/Documents/201008/wav-all/${f%.flac}.wav; done && \
cd /home/ubuntu/Documents/201006/audacity/xab-folder/flac-xab && \ [...]
for f in *.flac; do sox "$f" /home/ubuntu/Documents/201008/wav-all/${f%.flac}.wav; done

5. Start simon. Manage scenarios > New.
6. Simon > Vocabulary: Import lexicon-all.xml (this dictionary file contains 24000 German words). Import as active dictionary.
7. Grammar > Add sentence: Unknown
8. Training > Import training data:
/home/ubuntu/Documents/201008/prompts-all
/home/ubuntu/Documents/201008/wav-all
Importing 24,696 Files...
9. An error message appears:

Could not copy /home/ubuntu/Documents/201008/wav-all/mɪttʀaːgəndɐ.wav to /home/ubuntu/.kde/share/apps/simon/model/training.data//mɪttʀaːgəndɐ_9445_2010-08-25_19-12-34.wav. Please check if you have all the needed permissions.

What is the reason for this error message? I have to find out.

10. After solving some issues, the speech model is now available for download. It contains 24000 German words.

]]> http://spirit.blau.in/simon/2010/08/26/ralfs-german-speech-model-0-1-5/feed/ 0