I suggest you talk with some folks who test/monitor and certify equipment for the CE and/or UL labeling. You might start with:

Technology International Inc.
http://www.techintl.com

Washington Laboratories, LTD.
http://www.wll.com

David Baird
mrbaird@hotmail.com
 
Sr Controls Engineer

EET degree.

Journeyman Electrician.

I suppose you want to measure the sound generated by the DUT. In this case use a pick-up (direct contact, capacitive, inductuve or accoustic ) to synchronize
your tester and neglect anything nor coherent with
your sync.

<nbucska@pcperipherals.com>

I would try to use an anechoic chamber if possible that precludes background noise from entering. As to background seperating, I assume that you are going to use FFT analysis and that this might be a rotating machine like a motor or something. In this case, A motor will likely generate frequencies @ its fundamental and ball pass, 'on ball bearing, and cage frequencies. Unless the energy is one of four or five frequencies, you can safety ignore it. NI labview 6I would allow you to write bandpass filters for this also. As to sensor type, I would base this answer on frequency in question, very low frequencies I would use a contact capacitive accelerometer such as Silicon Designs model 2210, for higher frequencies, I quess I would use a piezo mic; but, I'm not sure what I'd use offhand.

I need to measure the sound of every product that comes out of the production. Some how I can not imagine connecting the contact mic and disconnecting it from every product. Do you think using a sound chamber is a better way to go?
If so , do you know of any app notes or book that will help me in designing a acoustic chamber properly?