Re: Speech recognition
Omer,
Thanks for your reply. Here are my thoughts on the links you sent:
> Speech Recognition HOWTO:
> http://www.faqs.org/docs/Linux-HOWTO/Speech-Recognition-HOWTO.html
As you mentioned, this site is six years out of date. It gives two
options for end-user speech recognition: XVoice and CVoiceControl.
XVoice has a sourceforge site and looks like a good speech recognition
user interface for commands and dictation but requires the proprietry
and no longer available IBM ViaVoice for Linux and so is no longer a
viable option (unless there is a way of making it work with another
underlying speech recognition system). The links to CVoiceControl are
dead.
> Speech Recognition for Linux:
> http://volker.dnsalias.net/linux/speechrec.html
This site mainly talks about how to get IBM ViaVoice for Linux working
and links to the How To site above.
> CVoiceControl - command and control for Linux:
> http://www.kiecza.net/daniel/linux/
Seems to be for issuing voice commands in the terminal. Potentially
useful as part of a speech recognition solution but lacks any
dictation facility or GUI command facility.
> Open Mind Speech:
> http://freespeech.sourceforge.net/
No updates since May 2000. I'm surprised the site is still active!
> The CMU Sphinx Group Open Source Speech Recognition Engines:
> http://cmusphinx.sourceforge.net/html/cmusphinx.php
This project develops open source speech recognition engines, but does
not appear to provide anything that could be easily downloaded and
installed to enhance desktop accessibility; there is no dictation or
desktop command program. To quote their homepage:
`Note however that Sphinx is not a final product. Those with a certain
level of expertise can achieve great results with the versions of
Sphinx available here, but a naive user will certainly need further
help. In other words, the software available here is not meant for
users with no experience in speech, but for expert users.'
> htk:
> http://htk.eng.cam.ac.uk/
The Hidden Markov Model Toolkit. This has something to do with speech
recognition, but I'm out of my depth here. This particular piece of
software appears to be owned by Microsoft and so probably does not fit
the Debian philosophy.
> simon:
> http://sourceforge.net/projects/speech2text/
This looks more like a useful integrated dictation and desktop control
system for end-users. It is still in alpha, but looks promising. It
seems to use HTK (see above), but as I am not sure how all the bits
fit together I do not know whether this is a problem in terms of
licensing.
> Julius:
> http://julius.sourceforge.jp/en_index.php
Like CMU Sphinx, this is speech recognition software without a user
interface to make it useful on the desktop. As you mentioned in your
email, it is used in Simon.
> MARF:
> http://marf.sourceforge.net/
This has something to do with speech recognition, but I honestly don't
understand what exactly it does. It's not a desktop application.
> Related Ubuntu project - VoxForge:
> http://www.voxforge.org/
> http://brainstorm.ubuntu.com/idea/4718/
Collection of open source audio in order to build acoustic models for
open source speech recognition. Would I be right in thinking that this
is the limiting factor for open source speech recognition on the
desktop? The VoxForge site implies this to be the case.
I am afraid I do not understand which components would be necessary
for an open source, native GNU/Linux speech recognition program(s)
that would allow dictation and the issuing of desktop-level commands.
There appears to be a fair amount of free software available, although
Simon appears to be the only project working towards an easy to use
desktop application. However, I wonder whether the more established
XVoice interface could be combined with any of the other programs
listed above.
As I said, I am very much out of my depth here. I wish I could be more help.
John
On 04/06/2008, Omer Zak <w1@zak.co.il> wrote:
> Hello John,
>
> There are lots of Free Software speech processing packages.
> The following is a random list of relevant resources.
> Let's evaluate and determine if any of those packages would be useful
> for you, and have it converted into a Debian package.
>
>
> Speech Recognition HOWTO:
> http://www.faqs.org/docs/Linux-HOWTO/Speech-Recognition-HOWTO.html
> Dates from April 2002, should be updated.
>
> Speech Recognition for Linux:
> http://volker.dnsalias.net/linux/speechrec.html
> Seems to be up to date as of 2005.
>
> CVoiceControl - command and control for Linux:
> http://www.kiecza.net/daniel/linux/
> Up to date as of 2002.
>
> Open Mind Speech: http://freespeech.sourceforge.net/
> The project seems to have died at 2000.
> The CMU Sphinx Group Open Source Speech Recognition Engines:
> http://cmusphinx.sourceforge.net/html/cmusphinx.php
>
> htk: http://htk.eng.cam.ac.uk/
>
> simon: http://sourceforge.net/projects/speech2text/
> Most recent news is from January 2008.
> Julius: http://julius.sourceforge.jp/en_index.php
> used by Simon, most recent software version was released at May 2008.
>
> MARF: http://marf.sourceforge.net/
> Seems to be active as of 2007.
>
> Related Ubuntu project - VoxForge:
> http://www.voxforge.org/
> http://brainstorm.ubuntu.com/idea/4718/
>
> The following are not speech recognition per se, but it may be good idea
> to ensure that there are Debian packages for all FOSS tools mentioned in
> the following:
>
> Free speech analysis software:
> http://www.personal.rdg.ac.uk/~llsroach/phon2/freespeech.htm
> Not directly related to speech recognition,
>
> Speech analysis software (Praat):
> http://homepage.ntu.edu.tw/~karchung/Phonetics%20II%20page%
> 20twentythree.htm
>
> --- Omer
>
>
> On Wed, 2008-06-04 at 17:42 +0100, John Hughes wrote:
>> Apologies in advance if this is not the right forum for this. I rely
>> on speech recognition a lot, but effective (or in fact, any) dictation
>> software is only available for Windows. Specifically, Dragon
>> NaturallySpeaking is the only working speech recognition program that
>> I have been able to find. I would like to go back to Debian, but at
>> the moment it is not a realistic option. What are the Debian Project's
>> thoughts on open source speech recognition for desktop computing? And
>> is there anything humble non-programmers with limited use of the
>> keyboard can do to help?
>
> --
> Philip Machanick: "caution: if you write code like this, immediately
> after you are fired the person assigned to maintaining your code after
> you leave will resign"
> My own blog is at http://www.zak.co.il/tddpirate/
>
> My opinions, as expressed in this E-mail message, are mine alone.
> They do not represent the official policy of any organization with which
> I may be affiliated in any way.
> WARNING TO SPAMMERS: at http://www.zak.co.il/spamwarning.html
>
>
> --
> To UNSUBSCRIBE, email to debian-accessibility-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmaster@lists.debian.org
>
>
Reply to: