James (JJ) Johnston (standing, right) speaks about the human hearing system at the PNW Section May 8 meeting at the Old Redmond Firehouse Teen Center, Redmond WA.
The PNW Section's other May meeting was held at the Old Redmond Firehouse Teen Center in Redmond, WA, featuring James (JJ) Johnston about "Why You Hear What You Hear."
13 AES member and 34 nonmembers attended. The Society for Sensible Explanations (Seattle Skeptics) and Skeptics meetup groups were both invited, and we had a good turnout of both.
After a welcome from PNW Chair Steve Turnidge, and crowd self introductions, PNW committeeperson Dan Mortensen announced another live Frank Laico interview program for Shoreline Community College; the next AES meeting with film score recordist Shawn Murphy; and the PNW elections.
Then James (JJ) Johnston, retired ex-Bell Labs Researcher and now DTS chief scientist, covered some basic ideas about the working of the human CNS (central nervous system) and hearing, with perhaps some insight into the pitfalls of gauging sound quality with humans. While certainly not everything about the workings of the human brain is fully understood, some basic understanding is available.
For our simplified picture, he separated the auditory system into 2 parts, the periphery (such things as HRTF (head related transfer function), cochlear analysis, and "partial loudness" after the cochlea), and the CNS. The inner ear processes sounds to partial loudness, which goes down the auditory nerve.
Then he spoke about the CNS processing, creating the need to switch between A-B sounds well within 200mS (with no clicks) or the ability to discern fine differences in loudness or timbre is reduced - important to know when testing small audio differences. Deeper in the CNS, partial loudness sensations are analyzed in mono and binaural. Lots of data is lost by now, and one's memory can last seconds or so. One's analysis can be strongly guided by learning, experience and cognition.
Then auditory objects can be committed to long term memory as even more data reduction occurs. The process can be entirely steered by attention, cognition, other stimuli. So, it's very easy to be swayed in what you thought you heard. If you think something might be different, you probably will think it is different.
Thus, the nature of the human hearing process dictates that care is needed to craft useful audio comparisons in Double Blind Tests:
-They must have a falsifiable nature (i.e. be able to distinguish between perception and an actual effect)
-They must isolate the subject from changes in other stimuli than audio (and make them comfortable)
-They must be time-proximate - no clicks or gaps when switching; user controlled changeover
-There must be test Controls - (positive, negative, & anchoring elements)
-There must be trained, comfortable listeners
In a nonblind test, if you are told it's different, refocusing will result in your hearing a difference, regardless. This is a property of cognition.
He then played an audio demo: a music excerpt with a measured 13.6dB S/N ratio with perceptual addition of noise; then the same S/N figure, via wideband noise. One sounds OK, the other awful; raw S/N numbers mean nothing.
Audio Demo 2 was borrowed from researcher Poppy Crum, who got it from someone else - an excerpt of backwards "masking" was played. Some hear a few words. The purported words were then shown while the backwards music was played again. Suddenly, everyone could understand the words, which is refocusing in action.
A break was held, with some door prize winners:
DTS ball caps: Rene Jaeger, Ray Wallace, Jonathan Christian, Tom Stiles
The meeting continued with visiting the websites of "interesting" audio products claiming "unusual" performance (at high prices), with spirited discussion of the products.
A special thanks to Ken Wong, Teen Program Director for the City of Redmond, who runs the Old Firehouse Teen Center.
