Hieroglyphs (aka sphinxDoc) Information

Disclaimer : Information provided here may not represent the standpoints of Carnigie Mellon University or CMU Sphinx Group.

What is the Hieroglyphs?

The Hieroglyphs is a set of comprehensive documentation in a book form.

Status of Hieroglyphs

(20070311) Draft 3, Revision II is uploaded. Added mentions of PocketSphinx in preface, introduction.

(20070309) Revision I is uploaded. Rewritten the preface and some minor changes in other chapters.

(20070306) It has been a while since I worked on Hieroglyphs. During the time I have officially left CMU and join a small startup called Scanscout. (Thanks for Prof. Rudnicky. He is very kind to support my account.) Scanscout is still very new so I become very busy and make Sphinx not my priority in these days. This is fine for sphinx 3.x, SphinxTrain and toolkit. They are in good hands of Dave and Evandro. Hieroglyph though, it just bugs me that I never wrap it up. That's why I decided to trim some chapters in second draft and build the third draft. Information there is appropriate either for 3.5, 3.6 or the developing 3.7. It's still quite messy and looks like draft. However, I hope that this information could be useful for anyone who is interested in using Sphinx. Btw, I love the cover of this draft. It is really a hieroglyph. :-)

(20070311) (3rd Draft) Hieroglyphs: Building Speech Application Using Sphinx and Related Resources

The 3rd Draft Rev 2 (pp. 321)

This includes

Chapter 1 License Information

Chapter 2 Introduction

Chapter 3 Software Installation (pre-3.7, so sphinx-base is not yet discussed)

Chapter 4 Building Speech Recognition System (75% completion)

Chapter 5 Sphinx Front-end (Correct up to sphinx 3.6, it doesn't cover the latest development of Dave's work on the front-end)

Chapter 6 General Software Description of SphinxTrain

Chapter 7 Acoustic Model Training (Currently empty)

Chapter 8 Language Model Training (mostly from Phil. Clarkson's web pages)

Chapter 9 Search Structures and Speed-up of the Speech Recognizer (BBI is not included.)

Chapter 10 Speaker Adaptation using Sphinx 3

Appendix A Command-line Information. (This part is mostly automatically generated, not very up-to-date.)

(1st Draft) Hieroglyphs: Building Speech Application Using Sphinx and Related Resources

(20040927) Update Chapter 1, 9 and Title/TOC . Publish the 1st revision of Chapter 2, 8 and 11.

Detail changes:

Chapter 2: finally publish after some rewording. A draft, I guess I want to change most of the content in future

Chapter 8: A version of cutting and pasting Ravi's documents and the paper written by me and Jahanzeb. Nothing very satisfactory.

Chapter 9: fix the names of the adaptation tools.

Chapter 11: added Yitao's live-mode APIs and quickly wrote something in terms of usage.

Appendix A.II: added Evandro's help and example strings for wave2feat. Tools description for all SphinxTrain tools are completed.

(20040810) Update Chapter 1, add Title/TOC page and Full SphinxTrain Command Line Information Page.

(20040807) Two sample chapters can now be viewed in the following links

(New! 2nd Revision 20040927) Title Page and Table of Contents

(New! 3nd Revision 20040927) Chapter I : License and use of Sphinx, SphinxTrain and CMU LM Toolkit

(New! 1st Revision 20040927) Chapter II : Introduction to Sphinx, SphinxTrain and CMU LM Toolkit

(New! 1st Revision 20040927) Chapter VIII : Search Structure and Speed-up of Sphinx's recognizers

(New ! 2nd Revision 20040927) Chapter IX: Speaker adaptation using Sphinx

(New! 1st Revision 20040927) Chapter XI: Development using Sphinx

(New ! 2nd Revision 20040927) Appendix A.2: Full SphinxTrain Command Line Information

They are mainly for developers. One warning, the speaker adaptation tools are still not stable so use it with discretion :-)

Draft of Hieroglyphs

(New!) 5th draft of the outline - I added an FAQ chapter and slightly modified Chapter 11.

4th draft of the outline - I added chapters on 1, speaker adaptation and 2) how to incorporate Sphinx in your development. We have immediate need for them to be done.

3rd draft of the outline - Many feedbacks say that v2 may be too ambitious. I agreed with them so I trimmed nearly half of the content and focus on just writing the backboone of the document.

2nd draft of the outline -Expansion from the 1st draft.

Many people complained to me there is not documentation in using Sphinx. Many more complained with other things else, this is my general response (in my faq page). I admit that lack of documentation is a problem. That is why we are starting to write one. Due to the limitation of time, we will first focus on writing a manual for sphinx 3.4 and upper.