work in progress

1- Basic software installation

sudo apt-get install build-essential autoconf libtool automake libasound2-dev python-dev subversion sox libsox-fmt-all


2- Getting and installing Sphinx3

Download:
sphinx3
sphinxbase

mkdir sphinx && cd sphinx
tar xzpf sphinx3...
tar xzpf sphinxbase...
cd sphinxbase
./autogen.sh --prefix=/usr/local
make && make install
cd sphinx3
./autogen.sh --prefix=/usr/local
make && make install
ldconfig
cd python
nano setup.py (MODIFY THE PATHS)
python setup.py install

3- Download the patch, python script and lexical

wget http://puredata.info/Members/patrick/sphinx2pd
tar xzpf sphinx2pd.tar.gz
cd sphinx2pd
pd sphinx2pd.pd

Make your own lexical and language modeling files (optional)
Upload your textfile containing your puredata commands here:
http://www.speech.cs.cmu.edu/tools/lmtool.html

Download the gzip’d tar file & unzip it to /dic
Dump in binary:

svn co https://cmusphinx.svn.sourceforge.net/svnroot/cmusphinx/trunk/share/lm3g2dmp
cd lm3g2dmp && make && cd ..
lm3g2dmp/lm3g2dmp dic/XXXX.sent.arpabo dic
mv dic/XXXX.sent.arpabo.DMP dic/pd.dmp
mv dic/XXXX.dic dic/pd.dic

What is the difference between MIC and LINE level?
Level refers to the relative strength of the signal and is measured in decibels. LINE level sources are much-amplified signals over MIC (microphone) level signals. Line level is usually between -10 to +4 dbm in strength while MIC levels are normally -60 dbm.

Line OUT signal voltage and impedance levels?
The line output “standard” designed to drive a load of 600 ohms or greater, at a mean signal level of 0.775V RMS. An exception exists in respect of compact disc players, where the output level is most commonly 2V RMS.

Preamplifier schematic
Thanks to Andy Collinson for sharing his circuit design.

How does it sound?
Electret microphone
Sensitivity: -35 to +4dB
Signal to noise ratio: 62dB
Recorded at 96000 hertz @ 24 bits

emc_kalimba.wav (right click and save)

A simple bridge between glchess and pure data: