Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
1998-04-17
2001-08-14
Korzuch, William (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C704S231000
Reexamination Certificate
active
06275797
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method for testing a communication network system and, more particularly, to using speech recognition to measure voice path quality in communication networks.
2. Description of the Related Art
Packet-based wide area networks provide communication channels that are used for transmitting voice data. A sender at one end of the network needs some way of telling whether the voice data is transmitted and intelligibly received at a destination point in the network. Thus, a need exists for testing and measuring the quality of the communication channel used for transmitting voice data over a packet based network.
Conventional test equipment used for testing telephone networks verify the integrity and continuity of telephone lines. For example, analog telephone lines are traditionally tested by using loop back systems that make a large number of measurements including D.C. voltage levels, D.C. voltage level offsets, A.C. impedances, frequency and phase responses, and the like. The loop back system then performs an extensive analysis of these measurements to determine whether the telephone line meets predetermined quality levels.
Digital telephone systems are tested by inserting a pseudo-random digital sequence at one access point in the telephone network and analyzing the received digital sequence for errors at a second access point. Neither of these test methods provide an objective measurement standard for evaluating the quality of voice signals transmitted over packet based data networks.
The primary concept underlying the design of traditional telephony networks requires certain minimum performance standards for each network segment or element to ensure adequate end-to-end voice quality. Wide Area Networks (“WANs”) and Local Area Networks (“LANs”), on the other hand, are designed based on obtaining the best performance or effort of each of its components or sub-systems with virtually no minimum performance requirements for the network segment or element.
A distributed network, such as a WAN or a LAN, includes multiple subnetworks connected together through different network processing nodes such as routers, switches, etc. In order to test the continuity of the communication link through the entire network, each processing element or node in the link is individually tested. Testing each network processing node is time consuming and often inadequate to test the quality of the voice link because during any given voice conversation, the voice packets from one user to another are routed through different paths depending on network congestion and other similar considerations. Also, the voice path for incoming data packets may be different than the voice path for outgoing data packets. Thus, the quality of the voice link in a single conversation can vary depending on the quality of the voice paths used to route the different data packets.
A distributed network can also include network processing equipment from a variety of manufacturers. Each equipment manufacturer may use different testing procedures for evaluating equipment performance. Thus, test results for network subsystems comprising several individual pieces of equipment may be disjointed making it difficult to evaluate the performance of the overall network or to identify poorly performing equipment.
A conventional method of testing voice quality in distributed networks involves one person simply establishing a communication link from a first location in the network to another person in a remote location. The two people determine whether the connection was made properly and then evaluate the quality of the voice signal at both ends of the link. This end-to-end testing method based on the statistical interpretation of human listeners is called Mean Opinion Score (“MOS”).
Manual testing of this type require the people conducting the tests to make subjective judgments regarding voice quality. Because evaluation of voice quality varies depending on the personal experience, training, health, etc. of the people evaluating the telephone system, subjective evaluation techniques are inherently unreliable. Moreover, subjective tests do not provide qualitative, numeric results or scores that can be easily evaluated by others or by computers.
Accordingly, a need remains for a method of measuring voice path quality in a communication networks which generates reproducible, objective, and easily evaluated results without human interaction.
SUMMARY OF THE INVENTION
Digitized voice signals are transmitted from a sending computer or terminal to a receiving computer or terminal at the other end of the communication link. The digitized voice signals are evaluated by a speech recognition engine that identifies speech segments in the received voice pattern and produces a textual form of the recognized speech. The resulting text file is compared with a reference text file to generate a voice path quality factor. The reference text file is a file generated by a speech recognition engine by processing a non-distorted common reference voice pattern. There can be more than one unique reference voice pattern with different voice properties. Each voice property, in turn, is differently susceptible to distortion from different distortion sources. Thus, by testing the voice path using different voice patterns, distortion patterns arise which can be used to identify the source of distortion in different network elements or segments.
The voice path quality factor is an objective indicator of transmission and processing quality of the communication link. The voice path quality factor is used to fine tune and condition the communication network under test. The voice path quality factor is also used to indicate a threshold level of operation during system level testing of the communication network. The voice path quality factor is particularly helpful for Internet Service Providers to test the voice path quality of distributed networks.
The foregoing and other objects, features and advantages of the invention will become more readily apparent from the following detailed description of a preferred embodiment which proceeds with reference to the accompanying drawings.
REFERENCES:
patent: 4771425 (1988-09-01), Baran et al.
patent: 4819228 (1989-04-01), Baran et al.
patent: 4903261 (1990-02-01), Baran et al.
patent: 5020058 (1991-05-01), Holden et al.
patent: 5313454 (1994-05-01), Bustini et al.
patent: 5359646 (1994-10-01), Johnson et al.
patent: 5572570 (1996-11-01), Kuenzig
patent: 5758027 (1998-05-01), Meyers et al.
patent: 5835565 (1998-10-01), Smith et al.
patent: 5848384 (1998-12-01), Hollier et al.
patent: 3708002A (1988-09-01), None
IBM Corporation “IBM Voice Type: IBM Speech Programmer's Guide, Introductory Section” Dec. 5, 1997, pp. 1-21.
Armstrong Angela
Cisco Technology Inc.
Korzuch William
Marger & Johnson & McCollom, P.C.
LandOfFree
Method and apparatus for measuring voice path quality by... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for measuring voice path quality by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for measuring voice path quality by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2496083