Meeting Topic: Audio, Radio, Acoustics and Signal Processing - the Way Forward
Speaker Name: James D. (JJ) Johnston
Meeting Location: Shoreline Community College, Shoreline, WA
For the April PNW meeting, James D. (JJ) Johnston gave a reprise of his AES 133rd Convention Richard Heyser Memorial Lecture on " Audio, Radio, Acoustics and Signal Processing - the Way Forward," plus a 5 channel recording demo. Attendance was about 23 AES members and 55 total.
Throughout his career, JJ noted that he was discouraged from studying audio, as it was "full of nonsense, and we didn't understand things very well." Fortunately, at Bell Labs, there was work in hearing and audio.
Referring to the masking work of Harvey Fletcher, JJ stressed that signal to noise ratio in and of itself is usually pointless, in deference to frequency selectivity of human hearing. He played some audio examples to demonstrate this. Even today, arguments about S/N and THD abound.
Preference (what you like to hear) is inviolate, he proclaimed, not what measures more accurately. He gave a modern view of human perception, with lots of feedback and loss of data as humans interpret things. Ergo, the need for double blind testing (DBT). He gave an anecdote about a faux "tube vs transistor" amplifier comparison test he tried in college, with a fake switch and only the transistor amp functioning. Some groups still showed a high preference for the "tube" amp.
So, your preference is not my preference. But under certain circumstances, preference is bad - when examining JUST the auditory system; when systems are so flexible that only with a falsifiable result lets you proceed with an investigation; and just anecdotes. One must be able to do a scientific study.
Thus, another cause of the divide/hostility between engineers and artists:
-the S/N experience teaches the artist to ignore the engineer
-the lack of DBT teaches the engineer to ignore the artist
Next he spoke about Bell Labs, Audio research and MP3.
Although discouraged from working on audio, some Bell researchers helped him around the barriers, eventually leading to his work on modern perceptual coders like MP3 and AAC.
He had inspiration from many of the minds at Bell Labs, and by 1979, he got his first perceptual codec lesson on "upward spread of masking" where lows mask highs, but highs don't mask lows. This was evident on solo female voice, and meant more work on perceptual coding.
In the meantime, work progressed on array microphones and a digital earphone, and computers were just becoming powerful enough now to do coding models in real time.
JJ played "The 13dB Miracle" where an example of music with a mere 13dB S/N with shaped noise sounded good despite the low S/N. And so:
-Perception does not respond to broadband SNR in any really useful fashion
-Perception integrates all senses - always consider it.
-Reproducing one point in a room accurately does not reproduce the soundfield in the original venue
Finally, he made some requests:
-Can we stop arguing with each other, talk, and develop some understanding among the engineers why the artistic side (mixers, etc.) do what they do?
-Can we have the artistic side stop with the "talk to the hand" treatment?
-Please, no more wideband SNR arguments. Puhleeeze!
Door prizes were given after the cookie break.
After the break, 5 channel demo recordings by Dr. Michael Matesky, Paul Hubert and JJ
were played, made with an array of 5 equiangular hypercardioids. This showed a method of recording and conveying perceptual cues for the room, although accuracy not always equal to preference, as the same recording with some processing made a more "normal" recording.
A 5 speaker listening area was set up in the room, and people could try the seats in the sweet area.