Hi Ralf,

>I would like to know how it is possible to make these two changes undone

>(how to revert these changes)?

Glad to see your learning Subversion!

You do this with the "merge" command:

Undoing Changes

An extremely common use for svn merge is to roll back a change that has already been committed. Suppose you're working away happily on a working copy of /calc/trunk, and you discover that the change made way back in revision 303, which changed integer.c, is completely wrong. It never should have been committed. You can use svn merge to “undo” the change in your working copy, and then commit the local modification to the repository. All you need to do is to specify a reverse--revision 303:302, or by an equivalent --change -303.) difference. (You can do this by specifying

$ svn merge -c -303 http://svn.example.com/repos/calc/trunk
--- Reverse-merging r303 into 'integer.c':
U    integer.c

$ svn status
 M     .
M      integer.c

$ svn diff
…
# verify that the change is removed
…

$ svn commit -m "Undoing change committed in r303."
Sending        integer.c
Transmitting file data .
Committed revision 350.

As we mentioned earlier, one way to think about a repository revision is as a specific changeset. By using the -r option, you can ask svn merge to apply a changeset, or a whole range of changesets, to your working copy. In our case of undoing a change, we're asking svn merge to apply changeset #303 to our working copy backward.

Note that this is different than Undoing Working Changes which applies to undoing changes your working copy (the copy of the Subversion repository that resides on your computer) - before they have been committed to the Subversion repository server.

Ken

Hi Ken! Thanks for the hyperlinks. Yes, I will try to learn more about svn. By the way, in my opinion we could need more technical writers who write more source code documentation. Regards, Ralf

Hi Ralf,

>we could need more technical writers who write more source code documentation

Which parts of the "source code" are you talking about?

Ken

Hello Ken! For example, there could be some documentation about these scripts from Timo (I haven't heard from him a few months, I hope that he continues the development of the German pronunciation dictionary).  Somebody could explain how to use these scripts, how they work, etc..

There is a lot of good software available, but it is so difficult to get involved.

Not everybody is an expert in the Perl programming language, and in Subversion.

For example, I think that the Subversion tutorial seems to be a pretty good one (I started reading this tutorial, this is just my first impression).  This book is for top-down learners.

In my opinion, the HTK book is for top-down learners, too.  But it is too complicated for normal people like me.  And I would be happy to have a shorter documentation that explains something about the abstract concept of HMM, and the connection to the HTK executables (or Sphinx or Julius).

So I need more documentation that gives a short explanation about the abstract concept (HMM).  And what is the function of the HTK executables (HCompV, HCopy, HDMan, HHEd, etc.)?  Which of those HTK programs are essential?  I don't want to know about the details (because it is too much for a beginner).  But I would like to understand the basics faster.

Recently, I saw that there is the Juliusbook (PDF).  And there is the HTK book.  What about a VoxForge book?

I am a top-down learner, but the HTK book with about 359 pages is too much for a beginner.  I would need an HTK book for starters with about 60 pages.  The problem for a beginner is: he wants to get an overview, but he doesn't want to read about the details.  What is essential for the fundamental understanding of the concept?

In my opinion, the graph theory is an abstract concept that is very easy understandable.  HMM is a bit more complex, and it needs a little bit more time to get involved.  But how are graph theory and HMM related to computational linguistics?  And how are they related to the VoxForge pronunciation dictionary?

Or there are very good German video tutorials about Simon available.  Isn't there someone who could make such videos in the English (Spanish, etc.) language?  This would be good for the marketing of Simon.  When more people would install Simon, they would learn about Julius, and HTK - and maybe support the ASR development.  So we have the chance to get more developers. And we need more developers because there is a lot of work to be done.  And with more marketing we could get more developers.  A good product sells better when there is the appropriate marketing.  And appropriate marketing means more technical documentation.

Or why does Simon need Julius, and HTK?  There is no explanation available.  The developer understands the concept.  There could be some technical writing.

Or why is it possible to download from VoxForge acoustic models that are compiled by Sphinx (Dutch, German, Russian)?  Why aren't they compiled with HTK?  What is the reason?  Some explanation would help.  And why are the English acoustic models made with HTK/Julius?

Regards, Ralf

Hi Ralf,

>Not everybody is an expert in the Perl programming language, and in Subversion.

Neither am I - I've had to learn all of this on my own with the VoxForge with lots of trial-and-error.  Most of my hands-on coding experience dates back to the late 1980s... in RPG and COBOL!

>In my opinion, the HTK book is for top-down learners, too.  But it is too complicated for normal people like me. 

I thought the HTK book is more for mathematicians... I have only really read the tutorial and could *not* understand most of the mathematics... I just use it for reference.

>And I would be happy to have a shorter documentation that explains something about the abstract concept of

>HMM, and the connection to the HTK executables (or Sphinx or Julius).

Me too... I don't understand the details of Hidden MarKov Models.  But I don't have to... I only need to know that they are statistical representations of sounds (for now at least...)

>And what is the function of the HTK executables (HCompV, HCopy, HDMan, HHEd, etc.)? 

The HTK book has a reference manual at the back

>Which of those HTK programs are essential?

Depends what you want to do... see the VoxForge Tutorial for the ones I used to create acoustic models...

>I don't want to know about the details (because it is too much for a beginner).  But I would like to understand the basics faster.

You gotta play with the tools...

>Recently, I saw that there is the Juliusbook (PDF).  And there is the HTK book.  What about a VoxForge book?

VoxForge Tutorial

>I am a top-down learner, but the HTK book with about 359 pages is too much for a beginner.  I would need an HTK book for

>starters with about 60 pages.  The problem for a beginner is: he wants to get an overview, but he doesn't want to read about the

>details.  What is essential for the fundamental understanding of the concept?

Unfortunately, there is no short-cut for this stuff... there are some books/tutorials you can buy that can help:

• Dan Jurafsky's SPEECH and LANGUAGE PROCESSING

• David Gelbart's links: Automatic Speech Recognition Theory and Algorithms

>But how are graph theory and HMM related to computational linguistics?  And how are they related to the VoxForge pronunciation dictionary?

er, dunno...

>Or there are very good German video tutorials about Simon available.  Isn't there someone who could make such videos in the

>English (Spanish, etc.) language? 

That's the beauty of open source - you see a need, and you fill it... the tools are free.  It just takes time and a bit of effort  :)

>Or why does Simon need Julius, and HTK?  There is no explanation available.

You'll have to talk to the Simon project

>Or why is it possible to download from VoxForge acoustic models that are compiled by Sphinx (Dutch, German, Russian)? 

Because Nickolay (nsh) was kind enough to create them and submit them to VoxForge

>Why aren't they compiled with HTK?  What is the reason?  Some explanation would help.

Because Nickolay uses Sphinx most of the time...

>  And why are the English acoustic models made with HTK/Julius?

Because when I started VoxForge, I did not know anything about speech recognition and HTK seemed to have better documentation... I have not gotten around to creating any in Sphinx.

In conclusion, I understand what you are saying about the lack of documentation on the VoxForge site, and that it would be nice to have more in the future.

Ken