Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Spreadsheet Program Accepting Continuous-Speech Input

IP.com Disclosure Number: IPCOM000117070D
Original Publication Date: 1995-Dec-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 4 page(s) / 144K

Publishing Venue

IBM

Related People

Britton, J: AUTHOR [+4]

Abstract

Disclosed are several techniques for providing a speech interface to a spreadsheet program.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 42% of the total text.

Spreadsheet Program Accepting Continuous-Speech Input

      Disclosed are several techniques for providing a speech
interface to a spreadsheet program.

      A spreadsheet program having this interface includes a
speaker-independent phonological model including all of the words
likely to be used with the program, even though only a small fraction
of this vocabulary is active for a particular user and for a
particular screen or context.  This data includes phonological data
for names commonly used to identify ranges, such as "sales."  This
model includes a facility to add new commands and names.

      Regional variations in speaking certain numbers and letters may
be accommodated by asking the user which accent he would like to use,
and by then loading baseforms and phonological data particularly for
that accent.  Regional variations include calling zero "naught" or
"zed."  Alternately, the user can be asked to speak several specific
numbers and letters as certain forms not used are deactivated, or the
system may start with a general language model, deleting some of the
possible pronunciations not used over time.  The system may start
with a particular phonetic model, choosing a better phonetic model by
checking the various models whenever a mistake occurs.  In general,
the speaker-independent model is adapted over time to the
pronunciation
of an individual user.

      A key to the effectiveness of the interface lies in an ability
to associate specific commands and cursor controls with pauses
exceeding a pause duration pre-established by the user.  A novice
user can set a pause parameter at a high level, so that he can, at
the end of an utterance, can see the numbers displayed on-screen with
the cursor left in position.  This adjustment allows the
inexperienced user to talk slowly, pausing between numbers and
digits.  On the other hand, an experienced user can set the pause
parameter at a low level, while associating a pause with a specific
action or sequence of actions, such as moving to the next cell in a
row or column.

      To facilitate the speech recognition process, names of cells of
the spreadsheet are constrained in one or more ways.  For example,
these names may be limited to those of cells which are visible on the
display screen, reducing the number of homophone-like combinations
which must be distinguished.  The possible cell names may also be
limited to cells currently active in the program.  The user may
identify, using a mouse or the keyboard, which columns and rows are
to be in the active vocabulary.  The cell names in the active
vocabulary of the system can also be limited to the cells identified
by the spreadsheet program, or by a template, as accepting or
requiring numeric data, or to only those cells used as input for
formula manipulation.  The spreadsheet program may include a variable
allowing cells to be identified uniquely as capable of speech input.

      Preferably, the user...