Meeting Topic: "Ensuring Accurate Playback & Analysis of Binaural Recordings"
Speaker Name: Dan Foley
Meeting Location: Devlin Hall, Boston College Campus, Chestnut Hill MA
Boston AES Meeting
On Tues. Sept. 14th, 2010, the Boston Section of the Audio Engineering Society featured Dan Foley with his presentation "Ensuring Accurate Playback & Analysis of Binaural Recordings" which was based on a paper presented last year at the 36th Audio Engineering Society Conference on Automotive Audio. Dan's Co-author of the original paper, Chris Struck, was also available from San Francisco via Skype to make comments and answer questions that came up. The presentation was meant to be a "Binaural Recording 101" so people who are new to the field can understand the issues of ensuring spatial accuracy of binaural recordings. The presentation covered Real Ear vs. Head-Related Transfer Function (HRTF) of manikin and factors that affect the recording and playback systems.
When measurements are taken from real people, one of the major factors in accuracy is based on the shape of the pinnae. It is important to note that if the shape of your pinnae were altered, you would have to completely re-learn spatial responses. When manikins are used, measurement are targeted for the population median so not completely accurate for anyone. This was supported by showing the wide range of responses of about 60 people against the manikin.
Dan outlined the details of taking measurements and described the type of mics uses in creating the measurements. He showed differences of the various mics used for binaural recording and gave key elements to issues that cause the accuracy of the recording to be affected. Measurement manikins are used for the telephony industry positioned at the eardrum locations. The ear canal is also a major variable in accuracy.
The Key factor for playback accuracy is to preserve all the timbral, temporal and spatial information.
Headphones are designed to replicate listening to loudspeakers in a living room environment. When listening using headphones, the information is not encoded with the torso and pinnae reflections of anybody, which needs to be taken into account. Headphones are equalized to have a response in a diffuse field for the "median" of the population. Dan and Chris answered questions posed about the methods manufacturers use (or don't use) to create the best response. Talking about type and styles of headphones.
Conclusions; If one wants to use a binaural manikin and judge sound quality, members of that panel should have pinnae that match the manikin as closely as possible. If recordings are made with probe mics on a person, you want to make sure there is minimal head movement during the recording. You have to account for the probe microphone's response characteristics and whether you have used the mic near the ear drum. You must make sure the noise floor is low enough to capture the appropriate sound levels. When using headphones you must take into account the diffuse field response of the headphones.
Dan Foley is the Principal of Foley & Associates, a Boston-based company specializing in technology and market development for the audio test, consumer, and professional products. He has worked in technical sales for Bruel & Kjaer and as Product Manager and Technical Marketing Specialist for Bose Corporation's Professional Products Group. He was also the Vice President of Sales and Marketing for Listen, Inc., which specializes in PC-based test and measurement systems used for production testing and research and development of electroacoustic products. Dan is a member of the IEEE Subcommittee on Telephone Instrument Testing and the Audio Engineering Society Working Groups on Digital Audio Measurement Techniques and Loudspeaker Measurement Techniques and has authored technical papers for the AES, and American Society of Mechanical Engineers.
Written By: Tony Schultz