Method and apparatus for measuring voice path quality by means of speech recognition

A method for testing voice path quality in a communication network, comprising:

• storing a plurality of sourced speech patterns that are not transmitted over the communication network;
• receiving a plurality of transmitted digitized voice packets, the plurality of digitized voice packets having been transmitted through one of a plurality of voice paths;
• converting the transmitted digitized voice packets into received speech patterns;
• feeding the received speech patterns into a voice recognition system for generating a corresponding plurality of recognized received digital text;
• feeding the sourced speech patterns into the voice recognition system for generating a corresponding plurality of recognized sourced digital text;
• comparing the recognized received digital text with the recognized sourced digital text;
• and
• generating a voice path quality factor for the selected voice path as a result of the comparison.

A system for testing a plurality of voice paths in a packet-based communication network, comprising:

• means for generating digitized voice test signals transmissible by the communication network through the plurality of voice paths;
• means for directly providing an original version of the digitized voice test signals to a plurality of first communication devices;
• means for transmitting the digitized voice test signals to one of the plurality of first communication devices through one of the voice paths;
• means for receiving a transmitted version of the digitized voice test signals at the one first communication device;
• means for feeding both the transmitted version of the digitized voice test signals and the original version of the digitized voice test signals to a voice recognition system for recognizing speech patterns in the transmitted version of the digitized voice test signals and in the original version of the digitized voice test signals at the one first communication device;
• means for comparing the recognized speech patterns in the original version of the digitized voice test signals with the recognized speech patterns in the transmitted version of the digitized voice test signals;
• and
• means for generating a voice path quality factor for the one voice path as a result of the comparison.

An electronic storage medium for storing software for testing voice path quality in a communication network, comprising:

• code for storing a plurality of sourced speech patterns that are not transmitted over the communication network;
• code for receiving a plurality of transmitted digitized voice packets containing the source speech patterns, the plurality of digitized voice packets having been transmitted through one of a plurality of voice paths;
• code for converting the digitized voice packets into received speech patterns;
• code for feeding the received speech patterns into a voice recognition system for generating a corresponding plurality of recognized received digital text;
• code for feeding the sourced speech patterns into the voice recognition system for generating a corresponding plurality of recognized sourced digital text;
• code for comparing the plurality of recognized received digital text with the plurality of recognized sourced digital text;
• and
• code for generating a voice path quality factor for the selected voice path as a result of the comparison.

A computer for analyzing the quality of voice signals in a network, comprising:

• a memory for storing original speech patterns that are not transmitted over the communication network;
• an input for receiving transmitted digitized voice packets, the plurality of digitized voice packets having been transmitted through any one of a plurality of voice paths;
• a decoder for converting the digitized voice packets into received speech patterns;
• a voice recognition engine for converting the received speech patterns into corresponding recognized received digital text and converting the original speech patterns into corresponding recognized original digital text;
• a processor for comparing the recognized received digital text with the recognized original digital text and generating a voice path quality factor according to how closely the recognized received digital text matches the recognized original digital text.