ftp.pasteur.org/FAQ//ftp-pasteur-org-FAQ.zip/FAQ/comp-speech-faq/part2

Internet Message Format | 1998-07-13 | 61KB

Path: senator-bedfellow.mit.edu!faqserv
From: [e-mail] (Andrew Hunt)
Newsgroups: comp.speech,comp.answers,news.answers
Subject: comp.speech Frequently Asked Questions - part 2/3
Supersedes: <comp-speech-faq/[e-mail]>
Followup-To: comp.speech
Date: 12 Jul 1998 12:00:23 GMT
Organization: Speech Applications Group, Sun Microsystems Laboratories
Lines: 1469
Approved: [e-mail]
Expires: 23 Aug 1998 12:00:04 GMT
Message-ID: <comp-speech-faq/[e-mail]>
References: <comp-speech-faq/[e-mail]>
Reply-To: [e-mail] (Andrew Hunt)
NNTP-Posting-Host: penguin-lust.mit.edu
Summary: Information on Speech Technology
X-Last-Updated: 1998/07/08
Originator: [e-mail]
Xref: senator-bedfellow.mit.edu comp.speech:18456 comp.answers:32122 news.answers:134643

Archive-name: comp-speech-faq/part2
Last-modified: 1998/07/06
URL: http://www.speech.su.oz.au/comp.speech/

                   COMP.SPEECH FAQ POSTING - PART 2/3


[Note: this document has been automatically extracted from a WWW site:
        http://www.speech.su.oz.au/comp.speech/
This may introduce some formatting errors.]


                        Signal Processing for Speech

                         comp.speech FAQ Section 2

          * SpeechLinks: Signal Processing for Speech
          * Q2.1: What sampling do I need for speech?
          * Q2.2: Finding the pitch of a speech signal
          * Q2.3: How do I find the start and end points of a speech
          signal?
          * Q2.4: Where can I find FFT software?
          * Q2.5: Signal processing in speech technology
          * Q2.6: Speech sampling and signal processing hardware
          * Q2.7: How do I convert to/from mu-law format?
          * Q2.8: Signal Processing Software


___________________________________________________________________________

               Q2.1: What sampling do I need for speech?

   For recorded speech to be understood by humans you need an 8kHz
   sampling rate or more and at least 8 bit sampling. This produces poor
   quality speech - but in can be understood.

   Improvements can be achieved by increasing the number of bits in
   sampling to 12bits or 16bits, or by using a non-linear encoding
   technique such as mu-law or A-law (see Q2.7). This improves the
   "signal-to-noise" ratio.

   Increasing the sampling rate above 8kHz, say to 10kHz, 16kHz or 20Khz,
   improves the frequency response: the higher the sampling frequency the
   better the high frequency content will be. A 16kHz sampling rate is a
   reasonable target for high quality speech recording and playback.

   When doing speech recognition you need to remember that the your
   computer is not as good as your ear so it will have trouble with poor
   quality sounds. The choice of an appropriate sampling setup depends
   very much on the speech recognition task and the amount of computer
   power available.


___________________________________________________________________________

               Q2.2: Finding the pitch of a speech signal

   This topic comes up regularly in the comp.dsp newsgroup. Question 2.5
   of the FAQ posting for comp.dsp gives a comprehensive list of
   references on the definition, perception and processing of pitch. The
   comp.dsp FAQ posting is posted regularly to the comp.dsp newsgroup,
   and is also available by ftp and on the WWW:

     * http://www.bdti.com/faq/dsp_faq.htm
     * ftp://rtfm.mit.edu/pub/usenet/comp.dsp/

   The following provide pitch tracking software:

     * Most of the speech processing environments listed in Q1.9
       including CSRE, ESPS, Kay Elemetrics Computer Speech Lab, OGI
       Speech Tools, Speech Filing System, Signalyze, Soundscope.


___________________________________________________________________________

         Q2.3: Finding start and end points of a speech signal

   End-point detection algorithms identify sections in an incoming audio
   signal that contain speech. Accurate end-pointing is a non-trivial
   task, however, reasonable behaviour can be obtained for inputs which
   contain only speech surrounded by silence (no other noises). Typical
   algorithms look at the energy or amplitude of the incoming signal and
   at the rate of "zero-crossings". A zero-crossing is where the audio
   signal changes from positive to negative or visa versa. When the
   energy and zero-crossings are at certain levels, it is reasonable to
   guess that there is speech. More detailed descriptions are provided in
   the papers cited below and in the documentation for the following
   software.

   End-point detection software is available from:

     * ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/tools/ep.1.0.tar.gz
     *
       ftp://ftp.isip.msstate.edu/pub/software/signal_detector/sigd_v2.2.t
       ar.gz

   Plenty of research papers have been presented on end-pointing. Try the
   following:

     * Rabiner LR, Sambur MR, "An Algorithm for Determining the Endpoints
       of Isolated Utterances", Bell System Technical Journal, Vol 54,
       No. 2, pp 297-315, 1975.
     * Drago, P.G. et al. "Digital Dynamic Speech Detectors." IEEE Trans
       on Communications, Vol 26, No 1, Jan 78, pp. 140-145.
     * Newman, W.C. "Detecting Speech with an Adapative Neural Network."
       Electronic Design. 22 March 1990.
     * Taboada. J et al "Explicit Estimation of Speech Boundaries" IEE
       Proc. Sci. Meas. Technol., Vol 141, No.3, May 1994, pp 153-159.


___________________________________________________________________________

                           Q2.4: FFT Software

   * Comprehensive list of FFT software
          Links to over 65 different pieces of one-dimensional FFT code.
          http://tjev.tel.etf.hr/josip/DSP/fft.html

   * FFT Software including optimised fft routines and mixed-radix
          algorithms
          ftp://usc.edu/pub/C-numanal/fft-stuff.tar.gz
          OR,
          ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/analysis/fft-stuff.
          tar.gz

   * mixfft03.zip: C-source for a very fast arbitrary N FFT routine
          The C-source is ShareWare: read the text file included in the
          package before using the FFT routine commercially.
          Jens J. Nielsen: [e-mail]
          Available from
          ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/analysis/mixfft03.z
          ip
          OR ftp://ftp.coast.net/simtel/msdos/c/mixfft03.zip

   * FFTW
          FFTW is a C subroutine library for computing the FFT in one or
          more dimensions. It is not limited to sizes that are powers of
          two, and includes real-complex and parallel transforms.
          Also on the FFTW web site are benchmarks comparing the
          performance and accuracy of many public-domain FFT
          implementations on a variety of platforms, as well as links to
          other sources of FFT code and information.
          Available from http://theory.lcs.mit.edu/~fftw
          Developed by Matteo Frigo and Steven G. Johnson:
          [e-mail]


___________________________________________________________________________

              Q2.5: Signal processing in speech technology

   This question is far to big to be answered in a FAQ posting. Here are
   some WWW resources and books which cover the area well.

  Tony Robinson's Course Notes

   Dr. Tony Robinson of the Engineering Dept of Cambridge University has
   put his Speech Analysis course notes on the web. The base page is
   http://svr-www.eng.cam.ac.uk/~ajr/SA95/. There is information on the
   following:

     * Sampling theory
     * Filter bank analysis
     * Short-term fourier analysis
     * Linear prediction analysis
     * Formant analysis and voicing analysis
     * Speech coding
     * and more....

  Joseph Picone's Course Notes

   Joseph Picone of the Institute for Signal and Information Processing
   (ISIP) at Mississippi State University has put two sets of course
   notes on the web:

   EE 4773/6773: Digital Signal Processing
          The course covers sampling, frequency analysis, z-transforms,
          filter design and more. The WWW site provides the syllabus,
          assignments, some source code data, exams, homework and
          solutions, lecture notes and more.

   EE 8993: Fundamentals of Speech Recognition
          The course covers background probability and
          phonetics/acoustics, speech signal analysis, dynamic
          programming, dynamic time warping, hidden Markov modelling,
          language modelling, neural networks, etc. The WWW sites
          provides the syllabus and lecture notes.

  Signal Processing Home page

   The Signal Processing Home page has information on a range of DSP
   issues. It includes references to a range of software and much more.
   http://tjev.tel.etf.hr/josip/DSP/sigproc.html

  Books and other References

   There are many good books which discuss signal processing for speech:

     * Digital processing of speech signals; L. R. Rabiner, R. W.
       Schafer. Englewood Cliffs; London: Prentice-Hall, 1978
     * Voice and Speech Processing; T. W. Parsons. New York; McGraw Hill
       1986
     * Computer Speech Processing; ed Frank Fallside, William A. Woods
       Englewood Cliffs: Prentice-Hall, c1985
     * Digital speech processing : speech coding, synthesis, and
       recognition edited by A. Nejat Ince; Kluwer Academic Publishers,
       Boston, c1992
     * Speech science and technology; edited by Shuzo Saito pub. Ohmsha,
       Tokyo, c1992
     * Speech analysis; edited by Ronald W. Schafer, John D. Markel, New
       York, IEEE Press, c1979
     * Applied Speech Technology Edited by: Ann Syrdal (AT&T Bell Labs,
       Holmdel, New Jersey), Raymond Bennett (Ameritech, Hoffman Estates,
       Illinois) and Steven Greenspan (AT&T Bell Labs, Murray Hill, New
       Jersey). Publisher: CRC Press.
     * Speech Communication: Human and Machine Douglas O'Shaughnessy,
       Addison Wesley series in Electrical Engineering: Digital Signal
       Processing, 1987.
     * Discrete-time processing of speech signals; John R Deller, John G
       Proakis, John H L Hansen; Macmillan 1993.
     * Signal processing of speech; F J Owens; Macmillan 1993.


___________________________________________________________________________

          Q2.6: Speech sampling and signal processing hardware

   In addition to the following information, have a look at the Audio
   File format document prepared by Guido van Rossum (see details in
   Section 1.8).

   Information is included on hardware for the following systems:

          * Macintosh Audio Hardware
          * PC Audio Hardware
          * Unix Audio Hardware

   Can anyone provide information for SGI, NeXT, other UNIX hardware and
   any other PC soundcards?



 Macintosh Audio Hardware - an overview

     * Description: ALL Macintosh computers come with the ability to play
       back sounds at any sample rate (sample rate conversion is done in
       software.) Older machines have 8 bit stereo output (hardware runs
       at 22254 samples/second). The newer machines have 16 bit stereo
       hardare running at 44100 samples/second.
       Most of the recent Macintosh computers come with sound input
       hardware. There are probably exceptions to this, but the older and
       some of the current low-end machines have 8 bit (linear) mono
       hardware running at 22254.54 samples/second. All of the PowerPC,
       AV, and the 500 series notebook computers come with 16 bit 44kHz
       stereo sampling hardware. They can also record at 22050
       samples/second. The sound manager implements an AGC (Automatic
       Gain Control) function for the 8 bit hardware. The drivers have a
       switch to turn off the AGC.
       There are a number of DSP vendors that support high quality audio.
       Generally this means quieter analog sections, and more IO formats
       (AES/IBU, for example). Try DigiDesign and Spectral Innovations.
       The software drivers for sound are described in "Inside Macintosh:
       Sound". If you want to see some sample code check out the sources
       for the Matlab "Sound and Image Toolbox". They can be found at

                ftp://ftp.apple.com/pub/malcolm/SoundAndImageToolbox.cpt.
                hqx

       Routines that play and record sounds using the toolbox are
       included (and interfaced to Matlab).



 PC Audio Hardware

   Note: new soundcards are becoming available all the time - the
   information below is definitely not up to date. Check out the
   following newsgroups for up-to-date information.

     * comp.sys.ibm.pc.soundcard
     * comp.sys.ibm.pc.soundcard.GUS
     * comp.sys.ibm.pc.soundcard.advocacy
     * comp.sys.ibm.pc.soundcard.games
     * comp.sys.ibm.pc.soundcard.misc
     * comp.sys.ibm.pc.soundcard.music
     * comp.sys.ibm.pc.soundcard.tech

   The Soundcard WWW Site is an excellent source of information:

     * http://www.wi.leidenuniv.nl/audio/

   An good source of programs and information for soundcards is SimTel:

     * http://www.acs.oakland.edu/oak/SimTel/win3/sound.html

   Additional information on PC soundcards is provided by the FAQ
   postings for the comp.sys.ibm.pc.soundcard.misc newsgroup. These are
   available by anonymous ftp from:
   ftp://rtfm.mit.edu/pub/usenet/comp.sys.ibm.pc.soundcard.misc/

     * Aria Soundcard FAQ
     * Aria Soundcard Support List
     * MIDI files software archives on the Internet
     * Turtle Beach sound cards FAQ



 Unix Audio Hardware

   Could someone please provide information on the audio capabilities of
   other Unix platforms?

    Sun standard audio port: SPARC I & II

     * Input and Output: 1 channel, 8 bit mu-law encoded, 8kHz sample
       rate. This provides telephone quality sampling.

    Sun DBRI audio port (SPARC 10 & 20)

     * Input and Output: Stereo (2 channels). 16-bit linear sampling.
       Multiple sample rates (48000, 44100, 37800, 32000, 22050, 18900,
       16000, 11025, 9600, 8000 Hz)

    Silicon Graphics Audio

   The Silicon Graphics audio Frequently Asked Questions (FAQ) is the
   best place to get information on SGI audio capabilities and
   programming. It provides information on connecting the audio output,
   using the DSP capabilities, controlling the audio output, programming,
   useful software and more. It is available from:

     * WWW: http://www-viz.tamu.edu/~sgi-faq/faq/html/audio/
     * News: comp.sys.sgi.misc
     * Ftp: ftp://viz.tamu.edu/pub/sgi/faq/

    Ariel Signal Processors

     * Platform: Various
     * Description: A range of signal I/O, A/D, D/A and DSP products are
       available. There are too many to list.
     * Contact: Ariel Corp.
       433 River Road, Highland Park, NJ 08904.
       Ph: 908-249-2900 Fax: [phone] DSP BBS: 908-249-2124


___________________________________________________________________________

             Q2.7: How do I convert to/from mu-law format?

   Mu-law coding is a form of compression for audio signals including
   speech. It is widely used in the telecommunications field because it
   improves the signal-to-noise ratio without increasing the amount of
   data. Typically, mu-law compressed speech is carried in 8-bit samples.
   It is a companding technqiue. That means that carries more information
   about the smaller signals than about larger signals.

   On SUN Sparc systems have a look in the directory /usr/demo/SOUND.
   Included are table lookup macros for ulaw conversions. [Note however
   that not all systems will have /usr/demo/SOUND installed as it is
   optional - see your system admin if it is missing.]

   OR, here is some sample conversion code in C.

/**
 ** Signal conversion routines for use with Sun4/60 audio chip
 **/

#include stdio.h

unsigned char linear2ulaw(/* int */);
int ulaw2linear(/* unsigned char */);

/*
** This routine converts from linear to ulaw
**
** Craig Reese: IDA/Supercomputing Research Center
** Joe Campbell: Department of Defense
** 29 September 1989
**
** References:
** 1) CCITT Recommendation G.711  (very difficult to follow)
** 2) "A New Digital Technique for Implementation of Any
**     Continuous PCM Companding Law," Villeret, Michel,
**     et al. 1973 IEEE Int. Conf. on Communications, Vol 1,
**     1973, pg. 11.12-11.17
** 3) MIL-STD-188-113,"Interoperability and Performance Standards
**     for Analog-to_Digital Conversion Techniques,"
**     17 February 1987
**
** Input: Signed 16 bit linear sample
** Output: 8 bit ulaw sample
*/

#define ZEROTRAP    /* turn on the trap as per the MIL-STD */
#define BIAS 0x84   /* define the add-in bias for 16 bit samples */
#define CLIP 32635

unsigned char
linear2ulaw(sample)
int sample; {
  static int exp_lut[256] = {0,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,
                             4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,
                             5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,
                             5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,
                             6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
                             6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
                             6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
                             6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
                             7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7};
  int sign, exponent, mantissa;
  unsigned char ulawbyte;

  /* Get the sample into sign-magnitude. */
  sign = (sample >> 8) & 0x80;          /* set aside the sign */
  if (sign != 0) sample = -sample;              /* get magnitude */
  if (sample > CLIP) sample = CLIP;             /* clip the magnitude */

  /* Convert from 16 bit linear to ulaw. */
  sample = sample + BIAS;
  exponent = exp_lut[(sample >> 7) & 0xFF];
  mantissa = (sample >> (exponent + 3)) & 0x0F;
  ulawbyte = ~(sign | (exponent << 4) | mantissa);
#ifdef ZEROTRAP
  if (ulawbyte == 0) ulawbyte = 0x02;   /* optional CCITT trap */
#endif

  return(ulawbyte);
}

/*
** This routine converts from ulaw to 16 bit linear.
**
** Craig Reese: IDA/Supercomputing Research Center
** 29 September 1989
**
** References:
** 1) CCITT Recommendation G.711  (very difficult to follow)
** 2) MIL-STD-188-113,"Interoperability and Performance Standards
**     for Analog-to_Digital Conversion Techniques,"
**     17 February 1987
**
** Input: 8 bit ulaw sample
** Output: signed 16 bit linear sample
*/

int
ulaw2linear(ulawbyte)
unsigned char ulawbyte;
{
  static int exp_lut[8] = {0,132,396,924,1980,4092,8316,16764};
  int sign, exponent, mantissa, sample;

  ulawbyte = ~ulawbyte;
  sign = (ulawbyte & 0x80);
  exponent = (ulawbyte >> 4) & 0x07;
  mantissa = ulawbyte & 0x0F;
  sample = exp_lut[exponent] + (mantissa << (exponent + 3));
  if (sign != 0) sample = -sample;

  return(sample);
}


___________________________________________________________________________

                    Q2.8: Signal Processing Software

   [Note: Question 1.9 lists speech laboratory environments and audio
   editors, many of which provide basic and advanced signal processing
   capabilities.]

Signal Processing Products

          * SigLib from Numerix Ltd.

On the Web

   The following sites provide lists of useful DSP software. Not all the
   software is directly applicable to speech processing.

   comp.dsp FAQ
          http://www.bdti.com/faq/dsp_faq.htm

   DSP Internet Resources
          http://www.eg3.com/
          http://www.eg3.com/dsp.htm

   Poynton's Digital Signal Processing Resource List
          http://www.inforamp.net/~poynton/Poynton-dsp.html

   WWW Pages Relating to Sound Computation
          http://datura.cerl.uiuc.edu/netstuff/sigsoundLinks.html

   Yahoo - Signal and Image Processing
          http://www.yahoo.com/Science/Engineering/Electrical_Engineering
          /Signal_and_Image_Processing/

   Sound Related Resources
          http://pscinfo.psc.edu/~geigel/menus/sound.html

   SPLIB: Signal Processing url LIBrary
          http://jazz.rice.edu/splib/

   Wavelet's Home Page
          http://www.mat.sbg.ac.at/~uhl/wav.html



SigLib from Numerix Ltd.

     * Platform: Windows, Unix and all major DSPs
     * Description: SigLib is an ANSI C Source DSP Library and includes
       functions for the following areas : spectrum analysis, windowing,
       filtering (fixed and adaptive coefficient), convolution,
       correlation, covariance, signal generation, statistical analysis,
       regression analysis, communications and modulation, digital
       effects, vectors processing, control, graphics and file I/O.
       Detailed product information and a description of the application
       of SigLib to speech processing is provided on the Numerix Ltd. WWW
       site.
     * Availability: A free demonstration of SigLib V2.0 is available
       from the Numerix Ltd. WWW site. Educational discount is available
       for SigLib.
     * Contact: Numerix Ltd.,
       157 Sileby Road, Barrow-on-Soar, Leics, LE12 8LW, UK.
       Phone/Fax : [phone]
       Email: [e-mail]
       WWW: http://www.numerix.co.uk/


___________________________________________________________________________

                       Speech Coding and Compression

                         comp.speech FAQ Section 3

          * SpeechLinks: Speech Coding
          * Q3.1: Speech compression techniques
          * Q3.2: Information on speech coding and compression
          * Q3.3: Speech Compression / Coding Software


___________________________________________________________________________

                  Q3.1: Speech compression techniques

   Provided by Tony Robinson:

   The aim of speech compression is to produce a compact representation
   of speech sounds such that when reconstructed it is perceived to be
   close to the original. The two main measures of closeness are
   intelligibility and naturalness.

   The standard reference point is toll quality speech, this is the same
   as what would be expected over a telephone line, for example, speech
   coded at 8 kHz using 8 bit ulaw coding and a maximum frequency of
   about 3.3 kHz. This is a bit rate of 64 kbps, and as such represents a
   compressed form over (say) 16 bit, 16 kHz speech which is the standard
   in speech recognition work.

   ulaw coding does not exploit the (normally large) sample to sample
   correlations found in speech. ADPCM is the next family of speech
   coding techniques, and does exploit this redundancy by using a simple
   linear filter to predict the next sample of speech. The resulting
   prediction error is typically quantised to 4 bits thus giving a bit
   rate of 32 kbps (see, for example, the software in Q3.3: 32 kbps
   ADPCM, G.711/721/723 Compression, shorten). The advantages of ADPCM
   are that is simple to implement and has very low delay.

   To obtain more compression specific properties of the speech signal
   must be modelling. The main assumption is known as the source filter
   model of speech production. This assumes that a source (voicing or
   fricative excitation) is passed through a filter (the vocal tract
   response) to produce the speech. The simplest implementation of this
   is known as a LPC synthesiser (e.g. LPC10e). At every frame the speech
   is analysed to compute the filter coefficients, the energy of the
   excitation, a voicing decision, and a pitch value if voiced. At the
   decoder a regular set of pulses for voiced speech or white noise for
   unvoiced speech is passed through the linear filter and multiplied by
   the gain to produce the speech. This is a very efficient system and
   typically produces speech coded at 1200-2400bps. With clever acoustic
   vector prediction this can be reduced to 300-600bps. The disadvantages
   are a loss of naturalness over most of the speech and occasionally a
   loss of intelligibility.

   The CELP family of coders compensates for the lack of quality of the
   simple LPC model by using more information in the excitation. Each of
   a set of codebook of excitation vectors is tried and the index of the
   one that best matches the original speech is transmitted. This results
   in an increase in the bit rate to typically 4800-9600bps. Most speech
   coding research is currently directed towards CELP coders. (See, for
   example, CELP 3.2a, a TMS implementation, a G.728 LD-CELP vocoder, and
   the L&H implementation.


___________________________________________________________________________

           Q3.2: Information on speech coding and compression

  Reference Books

   The following books cover speech coding/compression.

     * Douglas O'Shaughnessy, Speech Communication: Human and Machine,
       Addison Wesley series in Electrical Engineering: Digital Signal
       Processing, 1987.
     * Bishnu Atal in ed. Fallside, F. and W. Woods, ed. Computer Speech
       Processing. London: Prentice/Hall International, 1985. N. S.
       Jayant and P. Noll, Digital Coding of Waveforms, Prentice Hall,
       ISBN 0-13-211913-7 01, 1984.
     * W.B. Kleijn and K.K. Paliwal (Eds.), Speech Coding and Synthesis,
       Elsevier, Amsterdam, 1995.
       Contents, preface etc on the WWW:
       http://www.elsevier.nl/section/engtech/scs/menu.htm
     * Thomas P. Barnwell, Kambiz Nayebi and Craig H Richardson, Speech
       Coding: A Computer Laboratory Textbook, John Wiley and Sons Inc,
       1996.
     * Schuyler R Quackenbush, Tom P Barnwell III, Mark A Clements,
       Objective Measures of Speech Quality, Prentice-Hall, 1988.

   And the are good tutorial articles.

     * Makhoul, J. "Linear Prediction: A Tutorial Review." Proc. of the
       IEEE 63 (1975): 561 - 580.

  On the WWW

    comp.compression FAQ
          Includes a few questions and answers on the compression of
          speech.
          ftp://rtfm.mit.edu/pub/usenet/comp.compression/

    Tony Robinson's Speech Analysis Course
          A complete course on speech analysis, including some stuff on
          speech coding.
          http://svr-www.eng.cam.ac.uk/~ajr/SA95/
          http://svr-www.eng.cam.ac.uk/~ajr/SA95/node78.html

    ITU Coding Standards
          Members of the ITU (International Telecommunications Union) can
          obtain copies of the Series G Recommendations (including
          G.711/721/723/728) from the ITU WWW site (http://www.itu.ch/)
          and from http://www.itu.ch/itudoc/itu-t/rec/g/g700-799.html.

    Jason Woodard's Speech Coding Page
          Introduction to speech coding plus information on a series of
          speech coding standards.
          http://www-mobile.ecs.soton.ac.uk/speech_codecs/index.html

    WWW searchable online-bibiliography for Phonetics and Speech
          Technology
          Over 8000 entries provided by Institut fur Phonetik at Johann
          Wolfgang Goethe-Universitat Frankfurt.
          http://www.uni-frankfurt.de/~ifb/bib_engl.html

    Ciaran McElroy's Speech Coding Page
          Introduction to many types of speech coding.
          http://wwwdsp.ucd.ie/speech/tutorial/speech_coding/speech_tut.h
          tml

  Examples of speech coding

    Nam Phamdo's Speech Coding Demonstration
          Examples of ADPCM, LD-CELP, CELP, LPC10 and CELP coding and
          coding over a noisy channel.
          http://admii.arl.mil/~fsbrn/phamdo/speech_demo.html

    Phil Karn's Digital/Analog Voice Demo
          Examples of several speech coding systems.
          http://www.qualcomm.com/people/pkarn/voicedemo/


___________________________________________________________________________

               Q3.3: Speech Compression / Coding Software

   The following speech compression software is described in the FAQ.

          * 32 kbps ADPCM
          * Castleton Network Systems - G.729 Voice Coder
          * CELP 3.2a & LPC-10
          * 8 Kbit/s CELP on the TMS320C5x family of DSP chips
          * CyberVoice
          * Rockwell's DigiTalk
          * File format conversion
          * G.711/721/723 Compression
          * G.728 LD-CELP vocoder
          * G.728 Compression
          * GSM 06.10 Compression
          * Lernout & Hauspie Speech Coding (5 products)
          * Lernout & Hauspie Speech Coding SDK
          * MPEG Audio
          * shorten - a lossless compressor for speech signals
          * Sipro Lab Telecom Inc. Coding
          * Sonarc: Digital Audio Compression
          * StarAudio Compressor/Player
          * TrueSpeech from DSP Group
          * U.S.F.S. 1016 CELP vocoder for DSP56001
          * ToolVox from Voxware



32 kbps ADPCM

     * Platform: SGI and Sun Sparcs
     * Description: 32 kbps ADPCM C-source code (G.721 compatibility is
       uncertain)
     * Contact: Jack Jansen
     * Availablity: http://www.cwi.nl/ftp/audio/adpcm.shar



Castleton Network Systems - G.729 Voice Coder

     * Platform: TI TMS320C5x DSP
     * Description: G.729, also called CS-ACELP (Conjugate-Structure
       Algebraic Code Excited Linear Prediction), is a state-of-the-art
       voice compression ITU (International Telecommunications Union)
       standard that can be used in a wide range of applications
       including wireless communications, digital satellite systems,
       packetized speech and digital leased lines. G.729 provides 8000
       bits/s bandwidth for compressed speech at toll quality (equivalent
       to G.726 32 kbit/s ADPCM under clean channel condition). Also,
       G.729 has lower complexity and lower bit rate than G.728.
       The Castleton G.729 implementation provides a bit-exact
       implementation of the ITU standard on a single TI TMS320C5x DSP.
       The software is C callable and fully re-entrant, which allows easy
       interfacing and multi-channel capability. The encoder and decoder
       are fully independent, therefore, a DSP device can run a number of
       full-duplex or half-duplex channels. The coder and the decoder are
       able to operate under a real-time task switching kernel.
     * Cost and Availablity: Contact Castleton Network Systems.
     * Contact: Castleton Network Systems Corporation
       350 Terry Fox Drive, Kanata, Ontario, Canada K2K 2W5
       Ph: 613-591-8786, Fax: [phone]
       Email: [e-mail]
       WWW: http://www.castleton.com/



CELP 3.2a & LPC-10

     * Platform: Sun (the makefiles and source can be modified for other
       platforms)
     * Description: CELP is lossy compression technqiue. The US
       Department of Defences's Federal-Standard-1016 based 4800 bps code
       excited linear prediction voice coder version 3.2a (CELP 3.2a).
       Fortran and C simulation source codes.
     * Availability: By anonymous ftp from:
       ftp://ftp.super.org/pub/speech/celp_3.2a.tar.Z
       Or from the comp.speech ftp server
       ftp://svr-ftp.eng.cam.ac.uk/comp.speech/coding/celp_3.2a.tar.Z
       ftp://svr-ftp.eng.cam.ac.uk/comp.speech/coding/celp_3.2a.tar.gz
       LPC-10 Fortran source code is also available:
       ftp://ftp.super.org/pub/speech/lpc10-1.0.tar.gz
       Here is a modified LPC-10 release that includes ANSI C source:
       http://www.arl.wustl.edu/~jaf/lpc/
     * Documentation: The following articles describe the
       Federal-Standard-1016 4.8-kbps CELP coder:
          + Campbell, Joseph P. Jr., Thomas E. Tremain and Vanoy C.
            Welch, "The Federal Standard 1016 4800 bps CELP Voice Coder,"
            Digital Signal Processing, Academic Press, 1991, Vol. 1, No.
            3, p. 145-155.
          + Campbell, Joseph P. Jr., Thomas E. Tremain and Vanoy C.
            Welch, "The DoD 4.8 kbps Standard (Proposed Federal Standard
            1016)," in Advances in Speech Coding, ed. Atal, Cuperman and
            Gersho, Kluwer Academic Publishers, 1991, Chapter 12, p.
            121-133.
       The U.S. DoD's Federal-Standard-1015/NATO-STANAG-4198 based 2400
       bps linear prediction coder (LPC-10) was republished as a Federal
       Information Processing Standards Publication 137 (FIPS Pub 137).
       It is described in:
          + Thomas E. Tremain, "The Government Standard Linear Predictive
            Coding Algorithm: LPC-10," Speech Technology Magazine, April
            1982, p. 40-49.
       There is also a section about FS-1015 in the book:
          + Panos E. Papamichalis, Practical Approaches to Speech Coding,
            Prentice-Hall, 1987.
       The voicing classifier used in the enhanced LPC-10 (LPC-10e) is
       described in:
          + Campbell, Joseph P., Jr. and T. E. Tremain, "Voiced/Unvoiced
            Classification of Speech with Applications to the U.S.
            Government LPC-10E Algorithm," Proceedings of the IEEE Intl.
            Conf. on Acoustics, Speech, and Signal Processing, 1986, p.
            473-6.
     * Vendors:
       Realtime DSP code for FS-1015 and FS-1016 is sold by:
          + John DellaMorte, DSP Software Engineering
            165 Middlesex Tpk, Suite 206, Bedford, MA 01730, USA
            Ph: 1-617-275-3733 Fax: [phone]
            Email: [e-mail]
       DSP Software Engineering's FS-1016 code can run on a DSP
       Research's Tiger 30 (a PC board with a TMS320C3x and analog
       interface suited to development work).
          + DSP Research
            1095 E. Duane Ave, Sunnyvale, CA 94086, USA
            Ph: [phone] Fax: [phone]



8 Kbit/s CELP on the TMS320C5x family of DSP chips

     * Description: For low bandwidth transmission of voice, compact
       voice storage for archival purposes, low-cost digital answering
       machines and efficient storage for voice mail. Features :
          + near toll quality at 8 Kb/s.
          + Variable rate option with 1 Kb/s silence encoding.
          + Implemented on a fixed-point processor for lower system cost.
          + Attractive licensing scheme.
          + Future availability of 4 Kb/s.
          + Custom rates possible.
       Capacity :
          + Two half-duplex or one full duplex channels on the 20 MIPS
            'C5x (at 95% and 55% CPU utilization respectively).
          + Two full duplex channels on the 28.6 MIPS 'C5x (at 77% CPU
            utilization).
          + Requires 9 K-words program memory and 3 K-words data memory.
          + Decoding in real-time on a 486 class CPU.
     * Contact:

    CVI Inc.
    443 Vienna Cres. North Vancouver, BC, Canada V7N 3B3
    Tel: [phone] Fax: [phone]
    Email: [e-mail]



CyberVoice

     * Description: Cybernetics InfoTech, Inc. offers the following
       products
          + Telephone voice compression at 1.2, 2.4, 4.8 and 6.0 kbit/s
            with good-communications-quality to near-toll-quality coded
            voice;
          + Wideband voice (7-kHz bandwidth) compression at 16 kbit/s
            with near-original-quality coded voice;
          + Internet Voice E-mail software with voice editing,
            high-quality low-data-rate voice compression, fast/slow voice
            playback, and more.
     * Availablity: C code and Windows .DLL for telephone voice
       compression and wideband voice compression are available for
       licensing.
       Real-time DSP codes are under development.
       Voice E-mail software is available for purchase and download from
       the CyberVoice home page.
     * Contact: Cybernetics InfoTech, Inc.
       2 Professional Dr., #228, Gaithersburg, MD 20879
       WWW: http://www.cybit.com/
       E-mail: [e-mail]
       Fax: [phone]



Rockwell's DigiTalk

     * Description: The DigiTalk coder operates at a sampling rate of
       8KHz and transmits 223 bits of coded speech every 26ms, giving an
       overall bit rate of 8.577Kbps. The algorithm is based on
       analysis-by-synthesis predictive coding with vector-coded
       excitation, in which the excitation signal is optimized by
       minimizing the perceptually weighted error between the original
       and synthesized speech. More information and results of perceptual
       tests are available on the WWW.
     * Availablity: See the WWW page:
       http://www.nb.rockwell.com/ref/digitalk/



File format conversion

     * Platform: SUN OS?
     * Description: Conversion utility able to encode and decode between
       the the following formats: G.723, G.721, A-law, u-law and linear.
     * Availability: By anonymous ftp from

                 ftp://ftp.cwi.nl/pub/audio/ccitt-adpcm.tar.Z



G.711/721/723 Compression

     * Description:
          + G.711 : CCITT u-law and A-law compression
          + G.721 : CCITT 32 kbps ADPCM coder
          + G.723 : CCITT 24 kbps and 40 kbps ADPCM coders
     * Availability: By email to [e-mail], with
                GET ITU-3022
   as the *only* line in the body of the message.
       It is also available by anonymous ftp from:

                ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/coding/G711_G
                721_G723.tar.Z



G.728 LD-CELP vocoder

     * Platform: Analog Devices ADSP-2171
     * Description: Real-time, full-duplex G.728 LD-CELP vocoder that
       runs on a single Analog Devices ADSP-2171. Source and object code
       available for a one-time license fee.
     * Contact:

    Cole Erskine
    Analogical Systems
    299 California Avenue, Suite 120
    Palo Alto, CA 94306, USA
    Tel:[phone] FAX:[phone]
    email: [e-mail]



G.728 Compression

     * Description: G.728 low delay celp package written by Alex Zatsman
       of Analog Devices, Inc.
     * Availability: By anonymous ftp from

                 ftp://dspsun.eas.asu.edu/pub/speech/ldcelp.tgz



GSM 06.10 Compression

     * Platform: Unix; faster than real time on most Sun SPARCstations
     * Description: GSM 06.10 is a standardized lossy speech compression
       employed by most European wireless telephones. It uses RPE/LTP
       (residual pulse excitation/long term prediction) coding to
       compress frames of 160 13-bit samples (8 kHz sampling rate, i.e. a
       frame rate of 50 Hz) into 260 bits.
     * Contact: GSM 06.10 support and implementation
       [e-mail]_, [e-mail]
     * Availability: The following configurations are available be
       anonymous ftp:

                 gzip compression from Germany:
                ftp://ftp.cs.tu-berlin.de/pub/local/kbs/tubmik/gsm/gsm-1.
                0.7.tar.gz

                 MS-DOS compression from Germany:
                ftp://ftp.cs.tu-berlin.de/pub/local/kbs/tubmik/gsm/ddj/gs
                m-107.zip

                 MS-DOS compression from USA:
                ftp://ftp.mv.com/pub/ddj/1194.12/gsm-105.zip

     * Misc: The WWW site is

                http://www.cs.tu-berlin.de/~jutta/toast.html



Lernout & Hauspie Speech and Music Coding Product Range

     * Product name: L&H.smc650: 32kbps ADPCM Speech coding
          + Implementation of ADPCM 32 kbps based on CCITT G721 standard.
          + Estimated quality: 4.1 MOS (Mean Opinion Score)
          + Hardware Example: Analog Devices ADSP2101
          + Input / Output signal: A-Law or mu-Law PCM (64 kbps); Linear
            signal with up to 16 bits per sample; 8 kHz sampling rate
     * Product name: L&H.smc550: LD-CELP 16 kbps speech coding
          + Proprietary implementation of LD-CELP 16 kbps based on CCITT
            G728 standard.
          + Estimated quality: 4.0 MOS (Mean Opinion Score)
          + Hardware Example: Motorola 5600X
          + Input / Output signal: A-Law or mu-Law PCM (64 kbps); Linear
            signal with up to 16 bits per sample; 8 kHz sampling rate
     * Product name: L&H.smc450: 16-17.5 kbps speech coding
          + Estimated Quality: 3.9 MOS (Mean Opinion Score)
          + Hardware Examples: Analog Devices ADSP2101, Intel 486 DX2/66
            MHz
          + Input / Output Signal: A-Law or mu-Law PCM (64 kbps); Linear
            signal with up to 16 bits per sample; 8 kHz sampling rate.
     * Product name: L&H.smc350: 4.8-9.6 kbps speech coding
          + Proprietary CELP based software for compression rates of 4.8
            kbps to 9.6 kbps
          + Estimated Quality: 3.5 MOS (Mean Opinion Score)
          + Hardware Examples: AT&T DSP32C
          + Input / Output signal: A-Law or mu-Law PCM (64 kbps); Linear
            signal with up to 16 bits per sample; 8 kHz or 11.025kHz
            sampling rate.
     * Product name: L&H.smc250: 2.4 kbps speech coding
          + Combination of multi band excitation and code book excited
            linear prediction.
          + Estimated Quality: 3.0 MOS (Mean Opinion Score).
          + Hardware Examples: Intel 486 DX2/66 MHz, Analog Devices
            ADSP2101
          + Input signal: A-Law or mu-Law PCM (64 kbps); Linear signal
            with 12-15 bits per sample; 8 kHz sampling rate.
          + Output signal: A-Law or mu-Law PCM (64 kbps); Linear signal
            with 12-15 bits per sample; 8 kHz sampling rate.
     * See also: L&H Speech Coding SDK
     * More Information: On the WWW: http://www.lhs.com/coding.html
     * Cost: Unknown
     * Contact: Lernout and Hauspie Speech Products
       20 Mall Road, 4th Floor
       Burlington, MA 01803, USA
       Ph: [phone], Fax: [phone]
       Email: [e-mail]
       WWW: http://www.lhs.com/



Lernout & Hauspie Speech Coding SDK

     * Description: Windows based software development kit for
       integrating speech coding technology with Windows based PC
       applications.
     * Requirements: IBM-compatible 486 DX/33 MHz + 2MB RAM + MS DOS 5.0
       + MS Windows 3.1 (or higher) + Sound Blaster compatible sound
       board.
     * See also: L&H Speech Coding Products
     * More Information: On the WWW: http://www.lhs.com/coding.html
     * Cost: Unknown
     * Contact: Lernout and Hauspie Speech Products
       20 Mall Road, 4th Floor
       Burlington, MA 01803, USA
       Ph: [phone], Fax: [phone]
       Email: [e-mail]
       WWW: http://www.lhs.com/



MPEG Audio

   MPEG (Moving Pictures Experts Group) is a standard methods for
   compression and transmission of digital video and audio. Detailed FAQs
   and WWW sites are available for MPEG:

    MPEG Pointers and Resources
          http://www.mpeg.org/

    FAQ by Luigi: http://www.crs4.it/~luigi/MPEG/mpegfaq.html

    FAQ by Frank Gadegast
          http://www.powerweb.de/mpeg/mpegfaq/

    FAQ by by Chad Fogg
          http://www-plateau.cs.berkeley.edu/mpegfaq/MPEG-2-FAQ.html

    How to Install an MPEG Audio Player for your Web Navigator
          http://www.mpeg.org/index.html/MPEG-audio-player.html

MPEG Audio Software on the WWW

    Audio and Music Applications for Silicon Graphics Systems
          Lists 4 MPEG audio players for SGI machines.
          http://reality.sgi.com/employees/cook/audio.apps/public.html

    MPEG-1 Audio Layer 3 encoder, decoder and FAQ
          From the Fraunhofer Institute
          http://www.iis.fhg.de/departs/amm/layer3/index.html

    MPEG-2 Audio FAQ from Philips
          http://www.keymodules.philips.com/MD/mpeg/faqmpeg2.htm

    MPEG-1 and MPEG-2 audio software
          Universitaet Hannover
          ftp://ftp.tnt.uni-hannover.de/pub/MPEG/audio/

    MPEG-1 Audio Layer 1 &2 encoder - decoder
          Internet Underground Music Archive (IUMA)
          ftp://ftp.iuma.com/audio_utils/converters/source/

    Buddy Software Library: MPEG-1 Audio Layer 3 encoder and
          player
          http://www.buddy.org/softlib.html

    MPEG-1 Audio Layer 1 & 2 decoder and verifier at CCETT
          ftp://ftp.ccett.fr/pub/mpeg/audio_new/

    MPEG-2 Audio encoder and decoder at CCETT
          ftp://ftp.ccett.fr/pub/mpeg/mpeg2/

MPEG Audio - MetaSound

     * Platform: MS Windows/3.1 and Windows/95
     * Description: MetaSound is a partial MPEG-1 software decoder which
       is designed to work with hardware video decoders. It can reduce
       the hardware cost by eliminating the need for a hardware audio
       decoder. Currently, MetaSound has been successfully incorporated
       to work with three hardware video decoders. Features
          + Performance: For 486 DX4-100 machines or above, MetaSound can
            deliver FM quality (22 KHz) sound. For Pentium-90 or above
            machines, MetaSound requires 40% CPU bandwidth to deliver CD
            quality (44.1 KHz) sound.
          + Portability: it can take less than one month to port to new
            hardware video decoders.
          + CD standard supports including Video CD 1.0, Video CD 2.0,
            and CDI.
          + User interface with full set of functions: volume control,
            stop, pause, forward, backward, mute, resume, select the
            previous/next program track (Video CD 2.0), randomly select a
            program track (Video CD 2.0).
          + Error Recovery: can automatically skip error bitstreams.
     * Contact: Meta Media, Inc.
       F8, #10-1, Ho-Ping East Rd. Sec. 1, Taipei, Taiwan, R.O.C.
       Ph: 011-886-2-369-3330, Fax: [phone]
       Email: [e-mail]



shorten - a lossless compressor for speech signals

     * Platform: UNIX/DOS
     * Description: A fast waveform coder suitable for a speech and music
       signals in a wide variety of file formats. The degree of
       compression is adjustable from lossless to three bits a sample.
       16bit 16kHz speech generally attains 50% lossless compression and
       16:3 compression of CDROM quality speech is obtainable with only
       minor audiable degredation.
     * Availability: Anonymous ftp - UNIX and DOS versions

                ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/coding/shorte
                n.tar.gz

                ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/coding/shorte
                n.tar.Z

                ftp://svr-ftp.eng.cam.ac.uk/pub/comp.speech/coding/shorte
                n.zip



Sipro Lab Telecom Inc. Coding

     * Platform: Various processors
     * Description: Coding software for several International Standards
       plus two Proprietary standards.
       International Standards
         1. PCS 1900 (a 13 kbps codec, established as a North American
            PCS standard)
         2. Enhanced GSM (a 13 kbps codec)
         3. G.723 (8 kbps codec established as a multi-purpose
            international standard)
         4. G.729 (a dual-rate codec for the video phone market)
         5. G.729 Annex A (8 kbps codec made for Digital Simultaneous
            Voice & Data transmission in the modem industry).

       Proprietary Standards
         1. ACELP 8 v2.0 codec (flexible dual rate codec equipped with a
            VAD)
         2. ACELP 4.8 codec
     * Contact: Sipro Lab Telecom Inc.
       770, Chemin Lucerne, Ville Mont-Royal (Quebec), H3R 2H6 CANADA
       Ph: [phone], Fax: [phone]
       E-mail: [e-mail]
       WWW: http://www.sipro.com/



Sonarc: Digital Audio Compression

     * Platform: DOS and Windows
     * Description: Sonarc provides reversable, variable-rate compression
       of audio signals. Obtains compression ratio which averages about
       2:1. Supports monaural and stereo files, 8-bit and 16-bit files,
       and WAVE and VOC formats.
     * Availablity: Shareware by Richard P. Sprague
       Speech Compression
       P.O. Box 1785, Wilsonville, OR, 97070-1785, USA
       Ph: [phone]
       Email: [e-mail]



StarAudio Compressor/Player

     * Platform: Win95
     * Description: Using a time-domain process delivers lossless
       decompressed data. Processes any source of .wav file format, high
       quality 16-bit audio data at any sampling rate. Requires no
       special hardware and decompression speed is real-time on most
       486's and on any Pentium. The higher the sampling rate the higher
       the compression ratio; minimum compression of 4:1 for 11k data,
       and usually exceeding 7:1 for 44k data. Full bandwidth of signal
       is preserved with default compression options. Compression options
       allow increase of compression ratio further with a slight trade
       off in the reduction of the output quality. A decompression
       library is available for application development.
     * Demo: Download the shareware version of the program from the STR
       WWW site.
     * Misc: A technical paper is available in Word 6.0 format:
       ftp://ftp.speechtech.com/pub/speechtech/docs/audocw60.exe
     * Contact: Speech Technology Research Ltd.,
       Suite B - 1623 McKenzie Avenue, Victoria, B.C. V8N 1A6, Canada
       Ph: [phone]
       Email: [e-mail]
       WWW: http://www.speechtech.com/home/speechtech/



TrueSpeech from DSP Group

     * Description: TrueSpeech is a family of speech compression and
       decompression algorithms and software. It is designed for personal
       computers and personal communications devices. With the high
       compression ratios ranging from 15:1 to 27:1, TrueSpeech improves
       the storage and communications transmission of digital voice
       information and can be used in the integration of personal
       computers and telephones. TrueSpeech can be utilized in many
       products and applications such as:
          + Multimedia PCs
          + Sound cards and modems
          + Computer/telephony and teleconferencing
          + Voice mail systems and PBX systems
          + Wireless/cellular applications
          + Personal digital assistants
          + Games, Education
          + Video/cable and on-line services
       The TrueSpeech encoder is available for free in the Sound System
       of Windows 95 and Windows NT. The DSPG WWW pages have information
       on how to add TrueSpeech capability to your WWW pages.
     * Contact: DSP Group, Inc.
       3120 Scott Boulevard, Santa Clara, CA 95054-3317, USA
       Phone: [phone] Fax: [phone]
       Email: [e-mail]
       WWW: http://www.dspg.com/index.html



U.S.F.S. 1016 CELP vocoder for DSP56001

     * Platform: DSP56001
     * Description: Real-time U.S.F.S. 1016 CELP vocoder that runs on a
       single 27MHz Motorola DSP56001. Free demo software available for
       PC-56 and PC-56D. Source and object code available for a one-time
       license fee.
     * Contact:

    Cole Erskine
    Analogical Systems
    299 California Avenue, Suite 120
    Palo Alto, CA 94306, USA
    Tel:[phone] FAX:[phone]
    Email: [e-mail]



ToolVox from Voxware

     * Platform: Windows and soon available on Mac (in Beta now) and Unix
     * Description: ToolVox is a proprietary frequency domain speech
       coder. 11 KHz speech is coded to an average rate of between 5,000
       bits per second and 9,000 bps. Real-time compression algorithms
       available for 2,400 bps. 22 KHz playback, as well as a ultra low
       bit rate 8 KHz codec are coming soon. On playback, the time scale
       can be changed by a 5x factor, pitch can be modified over a 3
       octave range, and vocal personality can be modified using a
       tranformation function called VoiceFonts(tm).
     * Misc 1: A SDK for Windows is available.
     * Misc 2: Demo software is available from the Voxware Inc WWW page:
       http://www.voxware.com/
     * Price: Basic toolkit is $895 US. OEM and mass distribution
       licenses are separate. Ordering information is provided on the
       Voxware WWW server.
     * Contact:

    Voxware, Inc.
    Ph: [phone] Fax: [phone]
    Sale information: [e-mail]
    WWW: http://www.voxware.com/


___________________________________________________________________________

                        Natural Language Processing

                         comp.speech FAQ Section 4

   There is now a newsgroup specifically for Natural Language Processing;
   comp.ai.nat-lang. A FAQ posting is available for the group:

          ftp://rtfm.mit.edu/pub/usenet/comp.ai.nat-lang/Natural_Language
          _Processing_FAQ

   There is also a lot of useful information on Natural Language
   Processing in the comp.ai FAQ. That FAQ lists available software and
   useful references. It includes a substantial list of software,
   documentation and other info available by ftp.

   The FAQ has information on the following:

          * Q4.1: NLP References and Books
          * Q4.2: NLP Software


___________________________________________________________________________

                     Q4.1: NLP References and Books

   Take a look at the FAQ for the "comp.ai" newsgroup as it also includes
   some useful references.

     * James Allen: Natural Language Understanding, (Benjamin/Cummings
       Series in Computer Science) Menlo Park: Benjamin/Cummings
       Publishing Company, 1987.
          + This book consists of four parts: syntactic processing,
            semantic interpretation, context and world knowledge, and
            response generation.
     * G. Gazdar and C. Mellish, Natural Language Processing in Prolog,
       Addison Wesley, 1989
     * G. Gazdar and C. Mellish, Natural Language Processing in Lisp,
       Addison Wesley, 1989
     * G. Gazdar and C. Mellish, Natural Language Processing in Pop11,
       Addison Wesley, 1989
          + Emphasis on parsing, especially unification-based parsing,
            lots of details on the lexicon, feature propagation, etc.
            Fair coverage of semantic interpretation, inference in
            natural language processing, and pragmatics; much less
            extensive than in Allen's book, but more formal. There are
            three versions, one for each programming language listed
            above, with complete code.
     * Shapiro, Stuart C.: Encyclopedia of Artificial Intelligence Vol.1
       and 2. New York: John Wiley & Sons, 1990.
          + There are articles on the different areas of natural language
            processing which also give additional references.
     * Paris, Ce'cile L.; Swartout, William R.; Mann, William C.: Natural
       Language Generation in Artificial Intelligence and Computational
       Linguistics. Boston: Kluwer Academic Publishers, 1991.
          + The book describes the most current research developments in
            natural language generation and all aspects of the generation
            process are discussed. The book is comprised of three
            sections: one on text planning, one on lexical choice, and
            one on grammar.
     * Readings in Natural Language Processing, ed by B. Grosz, K. Sparck
       Jones and B. Webber, Morgan Kaufmann, 1986
          + A collection of classic papers on Natural Language
            Processing. Fairly complete at the time the book came out
            (1986) but now seriously out of date. Still useful for ATN's,
            etc.
     * Klaus K. Obermeier, Natural Language Processing Technologies in
       Artificial Intelligence: The Science and Industry Perspective,
       Ellis Horwood Ltd, John Wiley & Sons, Chichester, England, 1989.

   The following are extensive bibliographies related to NLP:

     * Computational Parsing : Syntactic Analysis, Semantic Analysis,
       Semantic Interpretation, Parsing Algorithms, Parsing Strategies :
       BIBLIOGRAPHY, by Conrad F. Sabourin 1994, 2 volumes, 1029p, ISBN
       2-921173-02-6, INFOLINGUA inc., P.O. Box 187 Snowdon, Montreal,
       H3X 3T4, Canada.
     * Computational Text Understanding : Natural Language Programming,
       Argument Analysis : BIBLIOGRAPHY, by Conrad F. Sabourin 1994,
       657p, ISBN 2-921173-06-9, INFOLINGUA inc., P.O. Box 187 Snowdon,
       Montreal, H3X 3T4, Canada.
       See also: http://gomer.mlink.net/infolingua.html
     * Computational Text Generation : Generation from data or Linguistic
       Structure, Text Planning, Sentence Generation, Explanation
       Generation : BIBLIOGRAPHY, by Conrad F. Sabourin with a survey
       article by Mark T. Maybury 1994, 649p, ISBN 2-921173-07-7,
       INFOLINGUA inc., P.O. Box 187 Snowdon, Montreal, H3X 3T4, Canada.
       See also: http://gomer.mlink.net/infolingua.html
     * Natural Language Processing : Interfaces to Databases, to Expert
       Systems, to Robots, to Operating Systems, and to
       Question-Answering Systems : BIBLIOGRAPHY, by Conrad F. Sabourin,
       1994, 2 volumes, 847p, ISBN 2-921173-08-5 INFOLINGUA inc., P.O.
       Box 187 Snowdon, Montreal, H3X 3T4, Canada
       See also: http://gomer.mlink.net/infolingua.html

Journals

   The major journals of the field are

     * Computational Linguistics and _Cognitive Science_ for the
       artificial intelligence aspects,
     * Cognition for the psychological aspects,
     * Language and _Linguistics and Philosophy_ and Linguistic Inquiry
       for the linguistic aspects.
     * Artificial Intelligence occasionally has papers on natural
       language processing.

Conferences

   The major NLP conferences are

     * ACL: held annually
     * COLING: held biannually

   Most AI conferences have a NLP track; AAAI, ECAI, IJCAI and the
   Cognitive Science Society conferences usually interesting for NLP.
   CUNY is an important psycholinguistic conference. Other conferences
   include NELS, the conference of the Chicago Linguistic Society (CLS),
   WCCFL, LSA, the Amsterdam Colloquium, and SALT.


___________________________________________________________________________

                           Q4.2: NLP Software

Natural Language Software Registry (NLSR) - NLP Tools

     * The Natural Language Software Registry is available from the
       German Research Institute for Artificial Intelligence (DFKI) in
       Saarbrucken. Its purpose is to facilitate the exchange and
       evaluation of natural language processing software within the
       research community. To this end, the NLSR is cataloging natural
       language software projects, both commercial and non- commercial.
       The new updated and enlarged version contains more than 100
       descriptions of natural processing software. Registry listings
       include:
          + speech signal processors, such as the Computerized Speech Lab
            (Kay Elemetrics)
          + morphological analyzers, such as PC-KIMMO (Summer Institute
            for Linguistics)
          + parsers, such as Alveytools (University of Edinburgh)
          + semantic and pragmatic analyzer, such as NLL (University of
            the Saarland, Germany)
          + generation programs, such as FUF (Ben Gurion University of
            the Negev)
          + knowledge representation systems, such as Rhet (University of
            Rochester)
          + multicomponent systems, such as ELU (ISSCO), PENMAN (ISI),
            Pundit (UNISYS), SNePS (SUNY Buffalo),
          + NLP-Tools, such as GULP (University of Georgia) or Linguist
            (Kansai Research Laboratory)
          + applications programs (misc.)
     * If you have developed a piece of software for natural language
       processing that other researchers might find useful, you can
       include it by returning the questionnaire available from the
       sources below.
     * ftp://ftp.dfki.uni-sb.de/pub/registry
     * e-mail: [e-mail]
     * Natural Language Software Registry
       Deutsches Forschungsinstitut fuer Kuenstliche Intelligenz (DFKI)
       Stuhlsatzenhausweg 3
       D-66123 Saarbruecken
       Germany
     * Other ftp sites are

        ftp://crlftp.nmsu.edu/pub/non-lexical/NL_Software_Registy

        ftp://dri.cornell.edu/pub/Natural_Language_Software_Registry

Part of Speech Tagger

     * Description: A rule-based part of speech tagger developed by Eric
       Brill.
     * Availability: The tagger software, about 10 descriptive papers and
       related data are available by anonymous ftp from
       ftp://ftp.cs.jhu.edu/pub/brill/


___________________________________________________________________________

   Copyright (c) 1993-6 by Andrew Hunt, all rights reserved.
   This FAQ may be posted to any USENET newsgroup, on-line service, or BBS as
   long as it is posted in its entirety and includes this copyright statement.
   This FAQ may not be distributed for financial gain.
   This FAQ may not be included in any collections or compilations
   without express permission from the author.



 ---

Andrew Hunt
Speech Applications Group
Sun Microsystems Laboratories       Ph:  [phone]
2 Elizabeth Drive, MS UCHL03-207    Fax: [phone]
Chelmsford, MA 01824, USA           Email: [e-mail]