Readme.txt for PTCWAVE.ZIP

PTCWAVE.ZIP contains files pertaining to "PTC: Perceptual Transform Codeing for Bandwidth Reduction of Speech in the Analog Domain," by Doug Smith, KF6DX. Part 1 appears in the May/June 2000 QEX, pp 9-12; Part 2 in Mar/Apr 2001 QEX, pp 9-17.

73, Bob, KU7G
1:40 PM Monday, February 26, 2001

Robert J. Schetgen, KU7G/1				Senior Assistant Technical Editor			
American Radio Relay League, 225 Main St, Newington, CT 06111-1494
ph: 860-594-0277 (direct)					fax: 860-594-0259	
e-mail: rschetgen@arrl.org  

QST Team  *  Hints & Kinks Editor   *  QEX Managing Editor							

Have you seen the new QEX? Wow! Check it out at http://www.arrl.org/qex/.

1.  Having unzipped the file, you have four files of interest.  They are:

     input.wav (original recording, sampling rate=31.250 kHz, BW=15.625 kHz)
     coded.wav (coded signal, sampling rate=7.8125 kHz, BW=3.90625 kHz)
     output.wav     (decoded signal, sampling rate=31.250 kHz, BW=15.625 kHz)
     readme.txt     (this file)

2.  In the coded sample, you may hear the pitch frequency of my voice as translated by the coder
to the 300-800 Hz territory.  It is the lowest-frequency tone you can hear, wavering in frequency
markedly with the voice.  You may also be able to tell that the different sub-bands aren't aligned
in time quite as well as they could be.  Clicks and pops are caused by the lack of windowing on
the input and output data.

3.  In the decoded sample, you may hear the effects of temporal and spectral smearing. 
Temporal smearing shows up as a slight slap-back echo effect.  Spectral smearing imparts a
certain roughness to the voice.  These things are the subjects of ongoing research.  I haven't
worked on it in a while, but I'm sure the codec can be further optimized.

4.  Those equipped with audio spectrum analysis software, such as SpectraPlus, will be able to
see the concentration of coded and decoded energy at discrete frequencies.  I think you can hear
that a bit, too.

5.  Some sound cards do not support the exact sampling rates I used.  They seem to select the
closest sampling rate available and that may alter the pitch of the voice a little. 

6.  A patent on this PTC process is pending.  Play the input and output samples on the air if you
like, but please don't play the coded sample there!  It may not be legal; I'm seeking clarification
on that.  Licensed hams may build and use the PTC process for non-commercial purposes.

7.  Your comments, please!

     73,

     Doug Smith, KF6DX
     PO Box 4074
     Sedona, AZ 86340
     dsmith@arrl.org
