5th Meeting - Open-Source Speech Recognition Initiative
Date: August 19-20, 2004
Time: 48 hours commencing at 00:01 AM GMT.
Location: OSSRI Listserv.
Attendance:
Directors:
Jessica Hekman
Ivan Uemlianin
Susan Cragin
Christophe Gerard
Dustin Wish
Volker Kuhlmann
Tristram Metcalf
Members:
Scott Carr
Eric Johansson
Visitors:
Willie Walker
Andrew MacGinitie
Agenda Items:
1. Adoption of the Agenda
Adopted by consensus.
2. Officer Reports.
A. Welcome of new members and listserv report. (Clerk.)
New Members:
None. (Andrew MacGinitie's application was not in time
for the agenda.)
Old Members:
Dennis A. Beckley
Scott Carr
Arthur Chan
Bruce Cyr
Alfred Dayton
Eric Johansson
Saravana Krishnamurthy
Mark Kyes
Turner Rentz III
Juan José García Rojo
Mark Ungewitter
Deleted from list:
Bruce Couper - no vail e-mail address.
Listserv Report:
As of this morning, there were 60 subscribers:
B. Incorporation Report (Clerk).
Two Changes to mission statement will be required.
1. ...to develop continuous-speech recognition software which supports
dictation and command-and-control, and runs on Open Source platforms
such as Linux.
IVAN - Yes.
CHRISTOPHE - Fine with me.
SUSAN - Yes.
DUSTIN - No Problems here. Thought I changed it already...
VOLKER - Sounds good.
TRIS - Keep open to phonetic text potentials for Deaf and HOH.
2. Make more general the anticipated licesing parameters. (Discussion?)
IVAN - How about:
- to release the code developed in this effort under an open-source license, such as the (L)GPL from the Free Software Foundation, or the BSD license.
or even
- to release the code developed in this effort under an OSI approved open-source license, for example, the GPL, LGPL, BSD, and MIT licenses.
http://www.opensource.org/licenses/
JESSICA - Or, even more generally, "to release the code developed in this effort under an open-source license"
DUSTIN - We should look at the GPL from GNU, but also like the MySQL project leave room for profiting from the products support or commercial licensing in the future. Still not sure about the Sphinx licensing GUI issue and how they license there project. I'm sure members of this group are much more acquainted with it than I.
WILLIE WALKER - Your lawyer would know best about this. I'm not a lawyer, so I'm not qualified to interpret licenses. Sphinx-4, however,
uses a BSD-style license, which (IMO) is a much better license than GPL. GPL tends to be quite incestuous and often chases people away from doing any commercial work with GPL related material.
IVAN - Just saying "open source license" seems best to me. It's for the mission statement - we don't even need to put "... as approved by". I agree with (?) that we should be free to chose the best license for each project when the project is established, or change it afterwards if we think it's necessary.
ERIC - licensing choices depends on what our goals are. If we are going to build something to give away and make sure all modifications stay publicly and openly accessible, then we should choose LGPL/GPL. I would suggest LGPL for libraries and anything else that could be linked into an application and GPL for self-contained or stand-alone components.
If we want to let anybody profit from our work and not give back their changes, then we should go with BSD.
If we want to make some money from our efforts whenever our work is incorporated into a commercial product but still leave it open and available for open source projects, then we will need to develop another form of license not unlike the sleepy cat license.
my bias is for the last option
TRIS - The SleepyCat License (no lice;) seems good as it appears to be a hybrid covering more & excluding less flexibility?
IVAN - Just "open source" would be the best. My only concern would be to define what we meant by an 'open-source license' maybe a reference to the open-source definition from the open source initiative would be enough, so:
- to release the code developed in this effort under an open-source license, as defined by the open-source initiative (http://www.opensource.org)
Re the discussions elsewhere about which license. I don't think the mission statement should go into this amount of detail. Ossri should be at liberty to release different projects under different licenses as we see fit.
3. Report from Goodwin Procter on the status of our incorporation.
SUSAN - Goodwin Procter did not respond to my request for information. I have heard nothing from them since Matthew Terry said he would file our papers by last Monday, August 16th.
We are still not on the Secretary of State's web site.
WILL MOVE DISCUSSION TO A THREAD.
C. Financial Reports (Treasurer).
3. Old Business / Committee and Task Force Reports:
#1 -- Standing Finance Committee
(Inactive. We have $0.)
#2 -- Fund-Raising Committee
(Ditto.)
#3 -- Product Marketing Task Force
(Ditto.)
#4 -- Product Engineering
(Ditto.)
#5 -- Web Site Committee
Champion: (Do we even have one?)
VOLKER - We need a volunteer to assemble more content. If we don't have any standing offers let's pick the first one we get... Ideally though
someone should be responsible for looking after/at the site to ensure
there isn't anything embarrassing on it.
Btw looks great so far.
#6 -- Technical Evaluation Task Force
Champion: Ivan Uemlianin. Members: Arthur Chan, Christophe Gerard.
Goal: to produce evaluation reports of currently available
free/open-source ASR software.
First Draft DUE.
IVAN: I have posted our current drafts to:
http://www.iau.ukfsn.org/iau/projects/devel/asr/ossri/index.html
Needless to say, the survey and the reports are works-in-progress.
I'm working on a similar report on the CMU SLM toolkit.
From discussion elsewhere, a report on Sphinx4 is needed.
THE REPORT WAS ACCEPTED WITHOUT COMMENT.
#7 -- Sourceforge Project (Task Force)
Champion: Scott Carr, Lead Developer. Member: Ed Suominen, Developer
(This was premature. See Proposal below, under New Business.)
4. New Business
A. OSSRI should have a wiki to gather ASR requirements. (Ivan proposal.)
The wiki should be moinmoin.
DISCUSS: Where should it live and who should feed it?
Committee? Task Force?
JESSICA - I have set one up at http://wiki.ossri.org/ossri/
(with help to redirect DNS :) . It is currently empty; the question of who should feed it remains. I am happy to have editorial control, and am also happy to pass that task on to someone else if they want it.
SCOTT CARR - The idea of a wiki is for everyone to contribute. There should be some checks and balances to make sure it doens't contain invalid information. I can help populate the Wiki.
IVAN - Jessica's looks good. Let's use it.
VOLKER - I understand the idea of a wiki being that anyone can edit every and everyone's previous deposits. The question of who should feed it is inherent: everyone.
TRIS - Sounds good.
THIS DISCUSSION MOVED AWAY FROM THE DIRECTORS MEETING TO THE THREAD "REQUIREMENTS PAGE ON WIKI"
B. Proposal - Establish Wine porting project as a Task Force, and
appoint Eric Johansson to head it. Eric will submit a report on the
status of the Task Force, and accept volunteers.
JESSICA - yes.
ERIC - This is acceptable provided we have enough volunteer help. The process of making NaturallySpeaking work on top of wine is not going to be simple and will require learning and communicating everything we do to make it work in order to make the process repeatable. This is not something one person can do alone in a reasonable time.
IVAN - Yes. (provided Eric is willing of course) Eric, you were approved to be an advocate for DNS on the wine project, yes? Could you tell us what that involves, and how the proposed ossri project would complement it?
CHRISTOPHE - I vote Yes.
DUSTIN - Sounds great.
VOLKER - I am in favour.
SUSAN - Yes.
C. Discussion - Would it be appropriate to have a Chief Technology
Officer for OSSRI? specifically, someone who would be focused keeping
the technology focus on the right track to getting us live as quickly
as possible. It might be appropriate for this to be a committee of two
or three people but not more than this. (Eric Johansson and Jessica
Hekman to lead discussion.)
JESSICA - OSSRI definitely seems to need direction. Eric and I both have strong opinions about that direction, but hopefully are able to listen to dissenting ideas. Does anyone object to our directing focus? I imagine we would propose tasks (of the code-writing sort) and marshall people to work on them. Eric, what do you think?
ERIC - I agree that in the big picture, it's setting direction etc.. At a more detailed level, I see the job of the CTO or technology committee as being:
proposing/collecting/refining projects
Prioritizing projects
putting funding targets on each project
working with funding committee to present credible discussion on the value of project and its duration.
handle day-to-day issues around intellectual property questions
creating content for public relations effort.
collecting updates on project progress.
It's a big task. Done right, it's an employee style position. when it's volunteer effort, you'll have to play second fiddle to making enough money to live on no matter who fills the role.
As for Jessica and myself being co-CTOs, I would be comfortable with that. We both have different interests and focus which should mix nicely in the position.
However at the end of the day, if we don't have volunteers to help us, progress will be exceedingly slow.
CHRISTOPHE - I agree. Any clear ownership of key areas (like Product Marketing requirement, Product Engineering requirement) should improve Ossri's ability to focus and to deliver. I have no preference for the form (Task force versus CTO).
DUSTIN - I also agree.
VOLKER - A good idea in principle, assuming there are more people who want to do the work than would be desirable to do the work efficiently. I'm not sure we're at that point yet. If someone does see a need for a single(or double)-person coordinator, then I am in favour.
TRIS - It's efficient to navigate final decisions from one control point, as long as all other eyes / ears have constant input to the helm.
SUB-ITEM - SPHINX GUI
DUSTIN - Also issue with adding the hacking of the Sphinx GUI to the list of projects should be addressed.
IVAN - I'm top tired to upload any serious SphinxTrain GUI requirements right now, but I will soon.
CHRISTOPHE - I like a lot the idea of the SphinxTrain GUI. We all know that hacking code is the last thing to do but sometimes, it is necessary to take some risk (like betting on Sphinx as the Speech vendor of choice) and do something with it to get some momentum and have people to rally around something concrete.
I personaly believe that we should go with Sphinx and develop a SDK as a
stepping stone - SphinxTrain GUI goes in that direction.
THIS DISCUSSION MOVED AWAY FROM THE DIRECTORS MEETING AND TO THE SPHINXTRAIN GUI THREAD.
D. 2005 OSSRI Annual Conference -- ? Schedule, dates, locations?
Susan to present suggestions.
VOLKER - A conference would be beneficial if it's running on some minimum scale. It's a huge effort to organise - the one I've just been to had an organisational lead time of 3 years. We would need something to present and talk about. My feeling is that it's a bit early, but we can start thinging about concepts. Are there any other speech recognition conferences running? That back-to-back idea is a good one.
TRIS - A Conference if overlapped or backed up to another existing symbiotic Conference could add energy to OSSRI that would otherwise be inaccessible.
SUSAN - Here are my thoughts. KISS. (Keep it Simple, Stupid.) If I can't make most of the arrangements myself, there's something wrong.
We really want to get a core group of people together. All we need to do that is about 20 reasonably-priced bedrooms and a meeting hall accessible in a minor way to the Boston airport. (Because more attendees are from Massachusetts than anywhere else, and we all have cars and can perform some shuttle service if needed.)
We don't need a WIFI network or a T-3 connection. Local dial-up is all right. Acoustics don't matter. We are meeting and talking, more than we are working.
The meeting should attempt to be informal and fun. Members should be able to bring spouses and families, if they wish. There should be recreational activities.
It would be nice to be able to bring in a conference telephone so one or two directors could participate from remote locations.
My theory is that the best places to do this are at the Maine, Cape Cod or Rhode Island shore, off-season. Lots of beach, cheap-ish rooms, good seafood, and funky atmosphere, so we can say we knew each other when...
Maine (Kittery to York) and Rhode Island have the edge because they are closer to Boston and more accessible via public transportation.
On Friday I visited the first possible location for the OSSRI first conference and retreat, tentatively planned for any 4 days March-May 2005. (The 1st and 4th days will be mostly "travel time.") The location was a beach resort on the Maine coast, accessible by train from Boston.
They have plenty of rooms, the prices are all right, there is take-out nearby, and they are right on the best stretch of the public beach. Kayak rentals are nearby. (Buy a wetsuit to swim in March.)
Some rooms are handicapped-accesible. The pool has a lift for the disabled.
http://www.wellsbeachmaine.com/
As of now they do not have sufficient meeting facilities for more than 20 people. We would have to take beach chairs out onto the sand or meet around the indoor pool.
Mark and I are moving to Rhode Island in the very early spring. Rhode Island may end up being the most convenient option. John Dowding is from Rhode Island and has volunteered some assistance with this, and we will work out more details later.
Since most of us will be giving talks to each other, start thinking about giving a 1/2 hour presentation and having your speech mostly written so we can put together a conference journal.
DISCUSSION WILL MOVE TO THREAD.
Adjourn.
Date: August 19-20, 2004
Time: 48 hours commencing at 00:01 AM GMT.
Location: OSSRI Listserv.
Attendance:
Directors:
Jessica Hekman
Ivan Uemlianin
Susan Cragin
Christophe Gerard
Dustin Wish
Volker Kuhlmann
Tristram Metcalf
Members:
Scott Carr
Eric Johansson
Visitors:
Willie Walker
Andrew MacGinitie
Agenda Items:
1. Adoption of the Agenda
Adopted by consensus.
2. Officer Reports.
A. Welcome of new members and listserv report. (Clerk.)
New Members:
None. (Andrew MacGinitie's application was not in time
for the agenda.)
Old Members:
Dennis A. Beckley
Scott Carr
Arthur Chan
Bruce Cyr
Alfred Dayton
Eric Johansson
Saravana Krishnamurthy
Mark Kyes
Turner Rentz III
Juan José García Rojo
Mark Ungewitter
Deleted from list:
Bruce Couper - no vail e-mail address.
Listserv Report:
As of this morning, there were 60 subscribers:
B. Incorporation Report (Clerk).
Two Changes to mission statement will be required.
1. ...to develop continuous-speech recognition software which supports
dictation and command-and-control, and runs on Open Source platforms
such as Linux.
IVAN - Yes.
CHRISTOPHE - Fine with me.
SUSAN - Yes.
DUSTIN - No Problems here. Thought I changed it already...
VOLKER - Sounds good.
TRIS - Keep open to phonetic text potentials for Deaf and HOH.
2. Make more general the anticipated licesing parameters. (Discussion?)
IVAN - How about:
- to release the code developed in this effort under an open-source license, such as the (L)GPL from the Free Software Foundation, or the BSD license.
or even
- to release the code developed in this effort under an OSI approved open-source license, for example, the GPL, LGPL, BSD, and MIT licenses.
http://www.opensource.org/licenses/
JESSICA - Or, even more generally, "to release the code developed in this effort under an open-source license"
DUSTIN - We should look at the GPL from GNU, but also like the MySQL project leave room for profiting from the products support or commercial licensing in the future. Still not sure about the Sphinx licensing GUI issue and how they license there project. I'm sure members of this group are much more acquainted with it than I.
WILLIE WALKER - Your lawyer would know best about this. I'm not a lawyer, so I'm not qualified to interpret licenses. Sphinx-4, however,
uses a BSD-style license, which (IMO) is a much better license than GPL. GPL tends to be quite incestuous and often chases people away from doing any commercial work with GPL related material.
IVAN - Just saying "open source license" seems best to me. It's for the mission statement - we don't even need to put "... as approved by". I agree with (?) that we should be free to chose the best license for each project when the project is established, or change it afterwards if we think it's necessary.
ERIC - licensing choices depends on what our goals are. If we are going to build something to give away and make sure all modifications stay publicly and openly accessible, then we should choose LGPL/GPL. I would suggest LGPL for libraries and anything else that could be linked into an application and GPL for self-contained or stand-alone components.
If we want to let anybody profit from our work and not give back their changes, then we should go with BSD.
If we want to make some money from our efforts whenever our work is incorporated into a commercial product but still leave it open and available for open source projects, then we will need to develop another form of license not unlike the sleepy cat license.
my bias is for the last option
TRIS - The SleepyCat License (no lice;) seems good as it appears to be a hybrid covering more & excluding less flexibility?
IVAN - Just "open source" would be the best. My only concern would be to define what we meant by an 'open-source license' maybe a reference to the open-source definition from the open source initiative would be enough, so:
- to release the code developed in this effort under an open-source license, as defined by the open-source initiative (http://www.opensource.org)
Re the discussions elsewhere about which license. I don't think the mission statement should go into this amount of detail. Ossri should be at liberty to release different projects under different licenses as we see fit.
3. Report from Goodwin Procter on the status of our incorporation.
SUSAN - Goodwin Procter did not respond to my request for information. I have heard nothing from them since Matthew Terry said he would file our papers by last Monday, August 16th.
We are still not on the Secretary of State's web site.
WILL MOVE DISCUSSION TO A THREAD.
C. Financial Reports (Treasurer).
3. Old Business / Committee and Task Force Reports:
#1 -- Standing Finance Committee
(Inactive. We have $0.)
#2 -- Fund-Raising Committee
(Ditto.)
#3 -- Product Marketing Task Force
(Ditto.)
#4 -- Product Engineering
(Ditto.)
#5 -- Web Site Committee
Champion: (Do we even have one?)
VOLKER - We need a volunteer to assemble more content. If we don't have any standing offers let's pick the first one we get... Ideally though
someone should be responsible for looking after/at the site to ensure
there isn't anything embarrassing on it.
Btw looks great so far.
#6 -- Technical Evaluation Task Force
Champion: Ivan Uemlianin. Members: Arthur Chan, Christophe Gerard.
Goal: to produce evaluation reports of currently available
free/open-source ASR software.
First Draft DUE.
IVAN: I have posted our current drafts to:
http://www.iau.ukfsn.org/iau/projects/devel/asr/ossri/index.html
Needless to say, the survey and the reports are works-in-progress.
I'm working on a similar report on the CMU SLM toolkit.
From discussion elsewhere, a report on Sphinx4 is needed.
THE REPORT WAS ACCEPTED WITHOUT COMMENT.
#7 -- Sourceforge Project (Task Force)
Champion: Scott Carr, Lead Developer. Member: Ed Suominen, Developer
(This was premature. See Proposal below, under New Business.)
4. New Business
A. OSSRI should have a wiki to gather ASR requirements. (Ivan proposal.)
The wiki should be moinmoin.
DISCUSS: Where should it live and who should feed it?
Committee? Task Force?
JESSICA - I have set one up at http://wiki.ossri.org/ossri/
(with help to redirect DNS :) . It is currently empty; the question of who should feed it remains. I am happy to have editorial control, and am also happy to pass that task on to someone else if they want it.
SCOTT CARR - The idea of a wiki is for everyone to contribute. There should be some checks and balances to make sure it doens't contain invalid information. I can help populate the Wiki.
IVAN - Jessica's looks good. Let's use it.
VOLKER - I understand the idea of a wiki being that anyone can edit every and everyone's previous deposits. The question of who should feed it is inherent: everyone.
TRIS - Sounds good.
THIS DISCUSSION MOVED AWAY FROM THE DIRECTORS MEETING TO THE THREAD "REQUIREMENTS PAGE ON WIKI"
B. Proposal - Establish Wine porting project as a Task Force, and
appoint Eric Johansson to head it. Eric will submit a report on the
status of the Task Force, and accept volunteers.
JESSICA - yes.
ERIC - This is acceptable provided we have enough volunteer help. The process of making NaturallySpeaking work on top of wine is not going to be simple and will require learning and communicating everything we do to make it work in order to make the process repeatable. This is not something one person can do alone in a reasonable time.
IVAN - Yes. (provided Eric is willing of course) Eric, you were approved to be an advocate for DNS on the wine project, yes? Could you tell us what that involves, and how the proposed ossri project would complement it?
CHRISTOPHE - I vote Yes.
DUSTIN - Sounds great.
VOLKER - I am in favour.
SUSAN - Yes.
C. Discussion - Would it be appropriate to have a Chief Technology
Officer for OSSRI? specifically, someone who would be focused keeping
the technology focus on the right track to getting us live as quickly
as possible. It might be appropriate for this to be a committee of two
or three people but not more than this. (Eric Johansson and Jessica
Hekman to lead discussion.)
JESSICA - OSSRI definitely seems to need direction. Eric and I both have strong opinions about that direction, but hopefully are able to listen to dissenting ideas. Does anyone object to our directing focus? I imagine we would propose tasks (of the code-writing sort) and marshall people to work on them. Eric, what do you think?
ERIC - I agree that in the big picture, it's setting direction etc.. At a more detailed level, I see the job of the CTO or technology committee as being:
proposing/collecting/refining projects
Prioritizing projects
putting funding targets on each project
working with funding committee to present credible discussion on the value of project and its duration.
handle day-to-day issues around intellectual property questions
creating content for public relations effort.
collecting updates on project progress.
It's a big task. Done right, it's an employee style position. when it's volunteer effort, you'll have to play second fiddle to making enough money to live on no matter who fills the role.
As for Jessica and myself being co-CTOs, I would be comfortable with that. We both have different interests and focus which should mix nicely in the position.
However at the end of the day, if we don't have volunteers to help us, progress will be exceedingly slow.
CHRISTOPHE - I agree. Any clear ownership of key areas (like Product Marketing requirement, Product Engineering requirement) should improve Ossri's ability to focus and to deliver. I have no preference for the form (Task force versus CTO).
DUSTIN - I also agree.
VOLKER - A good idea in principle, assuming there are more people who want to do the work than would be desirable to do the work efficiently. I'm not sure we're at that point yet. If someone does see a need for a single(or double)-person coordinator, then I am in favour.
TRIS - It's efficient to navigate final decisions from one control point, as long as all other eyes / ears have constant input to the helm.
SUB-ITEM - SPHINX GUI
DUSTIN - Also issue with adding the hacking of the Sphinx GUI to the list of projects should be addressed.
IVAN - I'm top tired to upload any serious SphinxTrain GUI requirements right now, but I will soon.
CHRISTOPHE - I like a lot the idea of the SphinxTrain GUI. We all know that hacking code is the last thing to do but sometimes, it is necessary to take some risk (like betting on Sphinx as the Speech vendor of choice) and do something with it to get some momentum and have people to rally around something concrete.
I personaly believe that we should go with Sphinx and develop a SDK as a
stepping stone - SphinxTrain GUI goes in that direction.
THIS DISCUSSION MOVED AWAY FROM THE DIRECTORS MEETING AND TO THE SPHINXTRAIN GUI THREAD.
D. 2005 OSSRI Annual Conference -- ? Schedule, dates, locations?
Susan to present suggestions.
VOLKER - A conference would be beneficial if it's running on some minimum scale. It's a huge effort to organise - the one I've just been to had an organisational lead time of 3 years. We would need something to present and talk about. My feeling is that it's a bit early, but we can start thinging about concepts. Are there any other speech recognition conferences running? That back-to-back idea is a good one.
TRIS - A Conference if overlapped or backed up to another existing symbiotic Conference could add energy to OSSRI that would otherwise be inaccessible.
SUSAN - Here are my thoughts. KISS. (Keep it Simple, Stupid.) If I can't make most of the arrangements myself, there's something wrong.
We really want to get a core group of people together. All we need to do that is about 20 reasonably-priced bedrooms and a meeting hall accessible in a minor way to the Boston airport. (Because more attendees are from Massachusetts than anywhere else, and we all have cars and can perform some shuttle service if needed.)
We don't need a WIFI network or a T-3 connection. Local dial-up is all right. Acoustics don't matter. We are meeting and talking, more than we are working.
The meeting should attempt to be informal and fun. Members should be able to bring spouses and families, if they wish. There should be recreational activities.
It would be nice to be able to bring in a conference telephone so one or two directors could participate from remote locations.
My theory is that the best places to do this are at the Maine, Cape Cod or Rhode Island shore, off-season. Lots of beach, cheap-ish rooms, good seafood, and funky atmosphere, so we can say we knew each other when...
Maine (Kittery to York) and Rhode Island have the edge because they are closer to Boston and more accessible via public transportation.
On Friday I visited the first possible location for the OSSRI first conference and retreat, tentatively planned for any 4 days March-May 2005. (The 1st and 4th days will be mostly "travel time.") The location was a beach resort on the Maine coast, accessible by train from Boston.
They have plenty of rooms, the prices are all right, there is take-out nearby, and they are right on the best stretch of the public beach. Kayak rentals are nearby. (Buy a wetsuit to swim in March.)
Some rooms are handicapped-accesible. The pool has a lift for the disabled.
http://www.wellsbeachmaine.com/
As of now they do not have sufficient meeting facilities for more than 20 people. We would have to take beach chairs out onto the sand or meet around the indoor pool.
Mark and I are moving to Rhode Island in the very early spring. Rhode Island may end up being the most convenient option. John Dowding is from Rhode Island and has volunteered some assistance with this, and we will work out more details later.
Since most of us will be giving talks to each other, start thinking about giving a 1/2 hour presentation and having your speech mostly written so we can put together a conference journal.
DISCUSSION WILL MOVE TO THREAD.
Adjourn.
posted by Susan at 8/25/2004 01:58:00 PM
0 Comments:
Post a Comment
<< Home