A Method for two phase speech key code analysis of flight crew voice conversations
Publication Date: 2015-Nov-30
The IP.com Prior Art Database
Speech analytics can assist in identifying strengths and weaknesses of processes and products Data such as flight crew voice conversations can be analyzed to provide insights on flight performance physical mental and cognitive state of the pilot the flight parameters as recognized by the pilot and actions taken by the pilot Current systems used for voice and speech analysis rely on post flight speech analysis and may not be able to associate flight parameters with speech during flight The proposed solution provides an easy to implement speech key code analysis for real time on board speech analysis and post flight on ground analysis
Page 01 of 7
A Method for two phase speech key-code analysis of flight crew voice conversations
Rohit S Prasad, Vasantha Selvi Paulraj
Speech analytics can assist in identifying strengths and weaknesses of processes and products. Data such as flight crew voice conversations can be analyzed to provide insights on flight performance, physical, mental and cognitive state of the pilot, the flight parameters as recognized by the pilot and actions taken by the pilot. Current systems used for voice and speech analysis rely on post flight speech analysis and may not be able to associate flight parameters with speech during flight. The proposed solution provides an easy to implement speech key-code analysis for real time on board speech analysis and post flight on ground analysis.
Data from cockpit such as flight crew speech, cockpit area voice, and other noise data provide opportunities for analysis of data. Currently voice data is recorded by tools such as Flight Data Recorder (FDR) and Cockpit Data Recorder (CDR). Several voice analysis based software are used for analyzing the voice data from the FDR and the CDR. Current voice data processing systems convert the speech received from the FDR/CDR into meaningful data using speech recognition and analysis tools. Although useful, if the voice data from FDR and CDR is not correlated with flight conditions, the conditions in the cockpit cannot be recreated or understood. Current systems used for voice data processing are complicated and require manual processing of stored voice data. With future cockpits expected to produce large volumes of data for advanced analytical functions, managing data and voice analysis becomes complex, memory intensive and time consuming. Hence, there is a need for simple, efficient and accurate systems to analyze voice data in both real time and for offline analysis.
Page 02 of 7
2. PROPOSED SOLUTION OF THE PROBLEM
The proposed solution provides a method for speech based analytics. A speech based key-code analysis is performed in two phases. The analysis is based on key-codes identified by monitoring of the flight crew voice conversations. The first phase includes on board real time analysis of captured key-codes from the raw speech. From the analyzed raw speech, several phrases and keywords related to aircraft, emergency and other categories can be extracted and converted to messages using key-codes. The key-codes can convert the speech to meaningful messages. The database contains pre-loaded specific key-codes.
Based on analysis of the flight crew conversations, and comparison of analyzed data, specific key-codes are identified. Each identified key-code is tagged with flight navigation parameters. The flight navigation parameters can include, but are not limited to, flight phase, GPS location, heading, altitude and attitude, cockpit atmospheric conditions, oxygen saturation, air-pressure, and autopilot status. The identified key codes can be used...