Audio files Quality Control

Required is an SQL database and a .NET application with interface to SQL database Scenario is a production line where operators listen to electric motors for a couple of seconds and judge them OK or NOK. There will be 2 pieces of software which may run concurrently on one PC or we may choose to run them on separate machines. 1) The 'Sample workstation' There will be an 8 (Mono) channel sound card or audio interface (Type TBD, suggestions welcome) and this sound card will be connected to 8 microphones. Another system via rs232 sends commands to start and stop recording as well as a serial number for a sample The PC then records records a sound sample from the appropriate sound channel untill commanded to stop (via rs232) Pc will then save the audio file (any lossles format) using the serial number as filenename Samples will be 1-5 seconds in duration. System needs to be ready to take a new sample 0.5 seconds after previous sample is finished All 8 channels will be triggered independently and so any number may be simultaneously recording at any given time 2) The ‘Operator workstation’ This software may run on a machine elsewhere on the network. or cuncurrently on the sample workstation machine The software will provide the main operator interface. Each operator will have a login and password (NOT a windows account). There will be an engineer level (again, not windows) for higher level functions Software will look for new sound samples generated by 1) Software will play sound samples and prompt operator for their decision. Operator will choose from 4 categories by clicking 1 of 4 buttons. The operator will be able to replay any sample before making a decision. The operator will be able to play any sample form a palette of reference samples to assist their decision. Operator’s decisions will be stored in the database with the serial number When all NEW samples have been listened to (operator can work faster than production line) , the system will present the operator with previous sound samples to be re-evaluated. The samples presented for re-analysis will include a mix of the following. A-Known BAD samples aka ‘Rabbits’ B-Known GOOD samples C-Earlier samples ( from that day) that have been evaluated by the current operator D-Earlier samples that have been evaluated by a different operator. The system will be able to generate reports based on the above to evaluate any/all operators performance. A) Report a particular operators performance: e.g “Today John correctly asseses 90% of Known BAD parts, Rejects 3% of Known GOOD parts, agrees with is own results 98% of the time, his assessment of other operators results concurs 95% of the time. B) Report (graph) any operators performance over time in the above 4 metrics to evaluate their progress/response to training. The system will continually determine and display metrics ABCD for the current operator so they have feedback on their performance. It should be possible to recall all data and the sound sample for a particular part based on its serial number