(no subject)

James P. Salsman (bovik@best.com)
Mon, 27 Jul 1998 08:47:04 -0700 (PDT)

Announcing the HUMAN PHONEME PROJECT --

"Free speech in the distance education classroom"

a non-profit scientific charitable endeavor

* * *

You are invited to contribute to the establishment of the
international Human Phoneme Project, which endeavors to make
scientific data comprising the recorded, transcribed,
aligned speech of children and adults useful to educational
projects accessible to all researchers and developers
involved with any project which may make use of such data.

Your donation can be used in one or more of the following ways:

1) to purchase the rights to the rigorously collected
speech data from the Linguistic Data Consortium, the LDC
source organizations, and the owners of any incidental
copyrights on the text of the spoken speech;

Ref.: http://www.ldc.upenn.edu/ -- Linguistic Data Consortium

2) to produce such high-quality labeled and aligned
speech information from useful and efficient available
sources (e.g., transcriptions of audio files posted on
public web sites), secure the rights to the resulting data,
and place it all in the public domain under the terms of the
GNU General Public License;

3) to lobby for the recognition of the Internet as the
world's "distance learning classroom" and thus a bona fide
classroom subject to the educational fair use exemptions of
the international Berne Copyright Treaty;

4) to benchmark the educational effectiveness of
commercially-produced and available speech recognition
engines and application programming interfaces (APIs) and
publish the results of such benchmarks in a manner that will
best encourage the commercial speech software systems
developers to address the concerns of the educational market.

Why is this necessary?

The Linguistic Data Consortium, formed by DARPA in the early
1990s at the University of Pennsylvania, has become
increasingly restrictive on the availability of their source
data to non-member organizations, membership costs and
qualifications have increased, and the data is being
published in smaller proportion to that produced by the
Consortium membership. Since speech recognition technology
has recently become commercially competitive, many LDC
members and commercial enterprises have been trying to
restrict access to the LDC data in order to prevent
competition. As more and more prior art for patented speech
systems makes restricting source data a commercially viable
means of stifling competition, the need to place useful
source data into the public domain becomes urgent.

How to help:

The international Human Phoneme Project is being
administered by Bovik Research Inst., a West Virginia
Nonprofit Organization as described in section 501(c)3 of
the U.S. Internal Revenue Code. In order to contribute to
the Project, please send your donation to the following
address with the form below.

Make checks or bank drafts in any currency payable to:

Human Phoneme Project

and please send to:

c/o James Salsman, Director
Bovik Research Inst.
121 Laurie Meadows Dr. # 457
San Mateo, CA 94403-5206
US

Please include the following form:

Donor Name: _______________________________________

Amount: ___________________________________________

Would you volunteer to help collect speech data? ___

Return address: ___________________________________

___________________________________

Country of citizenship: ___________________________

Email address: ____________________________________

Please use these funds for (check all that apply):

[ ] PURCHASE of LDC copyrights for public assignment
of the LDC speech corpora, transcriptions, derivative
works and adjunct copyrights under the terms of the GNU
project's General Public License. (When this project is
complete, excess funds which have not been earmarked for
other projects will be returned in proportion to the
donated amounts.)

[ ] PRODUCE additional human speech data either by direct
commission by contract for pay of the LDC or its member
organizations, or by the collection of additional data
through direct or indirect means, whichever is most
cost effective and productive of data useful for
educational applications.

[ ] LOBBY for recognition of educational "fair use"
exemptions to copyright for data used in an educational
context of any distance learning situation whether at
a traditional classroom or in a public venue.

[ ] BENCHMARK commercial speech recognition APIs to test
their suitability for educational uses; publish such
benchmark results to the customers of the producers.

Disclaimer: Lobbying funded by citizens of foreign countries
is performed under a Universal Covenant of Proxy affidavit
executed by a group of United States citizens which allows
for any political speech or donation by proxy under the
terms of articles 19, 20, and 21 of the Universal
Declaration of Human Rights, U.N. resolution 217 (A) III of
10 December 1948, to which all U.S. law is subordinate.

Sincere regards,
James Salsman,
Bovik Research Inst.