Browse Prior Art Database

Batch Processing of Audio Messages

IP.com Disclosure Number: IPCOM000115841D
Original Publication Date: 1995-Jun-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 35K

Publishing Venue

IBM

Related People

Sergeant, RP: AUTHOR

Abstract

Telephone callers frequently encounter voice processing devices that request they leave data such as Name, address, account number as an audio recording or message. Today these cannot be processed automatically by computers; they must be listened to and processed by humans.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 100% of the total text.

Batch Processing of Audio Messages

      Telephone callers frequently encounter voice processing devices
that request they leave data such as Name, address, account number as
an audio recording or message.  Today these cannot be processed
automatically by computers; they must be listened to and processed by
humans.

      The arrangement disclosed here takes advantage of the fact that
many current telephones such as ISDN (via User to User messages) or
ADSI equipped phones allow messages to be displayed at the phone
during the call.

      The voice processing device is provided with a facility to
enable its voice recognition algorithm to know the type of data it is
processing, e.g., a name or an address or an account number.  The
audio file then contains different types of data but is broken up
into different parts to facilitate voice recognition.

      Thus, during a call the display prompts the user with "SAY
NAME", "SAY ADDRESS", "SAY ACCOUNT NUMBER", etc so that a continuous
stream of voice data can be recorded (but with markers dropped in the
voice file to indicate start/stop of discrete pieces of voice data
such as NAME, ADDRESS etc).   When the file is processed later then
the recognition software will know precisely when to stop and start
processing for different data types resulting in fast processing with
higher success rates.

      Alternatively the markers partitioning the message may be held
in a separate annotation file.