Browse Prior Art Database

Tone Detection in the Presence of Speech by Projections onto Convex Sets

IP.com Disclosure Number: IPCOM000114632D
Original Publication Date: 1995-Jan-01
Included in the Prior Art Database: 2005-Mar-29
Document File: 4 page(s) / 163K

Publishing Venue

IBM

Related People

Nobakht, RA: AUTHOR

Abstract

Tone detection in the presence of speech has been attracting much attention in the past few years. This is due mainly to the wide availability of integrated digital systems such as fax, modem, and answering machines and their related call progress mechanisms and the necessity to discriminate among their signal characteristics. Disclosed is a technique for very reliable detection of tones in the presence of speech using Set Theoretic concepts. Set Theoretic estimation is governed by the notion of feasibility and produces solutions whose sole property is to be consistent with all information arising from the observed data and a priori knowledge. Each piece of information is associated with a set in the solution space and the intersection of these sets, the feasibility set, represents the acceptable solutions.

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

Tone Detection in the Presence of Speech by Projections onto Convex
Sets

      Tone detection in the presence of speech has been attracting
much attention in the past few years.  This is due mainly to the wide
availability of integrated digital systems such as fax, modem, and
answering machines and their related call progress mechanisms and the
necessity to discriminate among their signal characteristics.
Disclosed is a technique for very reliable detection of tones in the
presence of speech using Set Theoretic concepts.  Set Theoretic
estimation is governed by the notion of feasibility and produces
solutions whose sole property is to be consistent with all
information arising from the observed data and a priori knowledge.
Each piece of information is associated with a set in the solution
space and the intersection of these sets, the feasibility set,
represents the acceptable solutions.  By making direct projections
onto these sets, the feasibility set or the acceptable solution is
obtained.  There are many projection methods which can be used to
obtain the feasibility set.  The practical use of the set theoretic
framework stems from the existence of efficient techniques for
finding these solutions.

      In most problems, information to be used in determining the
acceptability of a proposed solution can be classified into three
groups, namely information about the solution, information about the
system, and the information about external factors.  Information
about the solution represents our direct knowledge about the
properties of the result and it explicitly defines the acceptability
of a proposed solution.  For the problem of detecting tones in the
presence of speech, these information could contain the frequency of
the tone(s) being detected, the duration of time required for the
detection, and the range of the frequency offset or deviation of the
tone(s).

      Consider an input signal x(n) consisting of certain tone(s)
mixed with speech and noise.  For a given tone frequency fc and a
sampling rate fs the over sampling rate N can be defined as fs/fc.
Based on the received sampled data, the initial set can be defined by
equation (1).  Letting M=jN based on this initial set, the following
sets can be defined by equation (2).  Each Si will be called a
property set.  Thus, Si is the set of all estimates that are
consistent with all the information available about the problem
(information arising from the data and a priori knowledge).  The
subset of the solution space of objects consistent with all available
information is the feasibility set, defined by equation (3).  S will
be called the solution set.  Any point in S will be called a set
theoretic estimate.  The projection of a point a_{l,i} in Si onto
S_{i+1} is any point a_{p,i+1} in S_{i+1} such that equation (4) is
satisfied.

      Such a point is also called a best approximation of a_{l,i} by
a point in S_{i+1}.  If the solution space is a Hilbert...