Subject: Re: AUDITORY Digest - 5 Apr 2002 to 7 Apr 2002 (#2002-50) From: "Watson, Charles S" <watson(at)INDIANA.EDU> Date: Mon, 8 Apr 2002 09:40:05 -0500Dr. Mercuri, Yes, there have been a number of studies of Morse code learning, but still one of the most interesting is: W. L. Bryan and N. Harter (1897) Studies in the physiology and psychology of telegraphic language. Psych. Rev. 4, 27-53, and also 345-375. This study has been cited for the apparent "plateau's" in the learning functions, giving rise to the hypothesis that operators used a single-letter transcription strategy until they could no longer improve their rates, when they learned to listen for larger patterns. While statistically those plateau's seem to be more in the eye of the beholder, their data remain remarkable for the performance changes over many months of 5-6 day per week training. Especially interesting is that they did not show any clear evidence of asymptotic reception rates, after months of training. Since we have little evidence with which to compare it, this leaves us wondering what the upper limit actually is. Chuck Watson Indiana University PS Incidentally, Bryan was one of the founders of experimental psychology in U.S., having been trained in Germany in Wilhelm Wundt's lab. He began our psychology department here at IU, and later served as the university president for some 30 years. csw -----Original Message----- From: Automatic digest processor [mailto:LISTSERV(at)LISTS.MCGILL.CA] Sent: Sunday, April 07, 2002 11:03 PM To: Recipients of AUDITORY digests Subject: AUDITORY Digest - 5 Apr 2002 to 7 Apr 2002 (#2002-50) There are 2 messages totalling 79 lines in this issue. Topics of the day: 1. maximum 'tatum' speed perception (2) ---------------------------------------------------------------------- Date: Sun, 7 Apr 2002 21:03:08 -0500 From: beauchamp james w <j-beauch(at)UX1.CSO.UIUC.EDU> Subject: Re: maximum 'tatum' speed perception > On April 3, 2002 Brian Whitman <bwhitman(at)MEDIA.MIT.EDU> wrote: > Has there been any human-testing work on the maximum amount of discrete > musical events a person can identify in a second? At what point should a > machine listener (who obviously could do better) stop perceiving rapidly > played notes or percussives? I think this would be difficult to test, but I believe Milton Babbitt did some informal testing in the 60's and 70's. You could do some pattern matching, but I don't know that that would prove the listener "heard" all or a certain percentage of the actual notes. On the other hand, some musicians are able to take dictation of extremely fast passages, e.g., a lot of fast bebop jazz has been transcribed. However, transcribers can employ tricks such as slowing down the sound with a variable speed recorder or repeating a passage over and over again, until he/she "gets it right". Pitch detector programs could also be employed, and in many cases they work very well. > At extremely high LFs (20Hz) we'll start perceiving tones, and in > playing-around testing on my computer I could hear quarter note snare > drums up to 800bpm (so, about 13Hz). Is this a function of musical > training, the sounds being played, or mechanics of the ear? Probably all three. My personal impression is that our ears have evolved to decode in some sense the fastest things that humans can play, which probably tops out around 20 pulses per second. But perhaps this varies with fundamental frequency of the source, since it takes less time (shorter duration) to establish a definite pitch at higher frequencies than at lower ones. Listening experience definitely improves ability to appreciate and recognise virtuostic performances and recognize patterns within these performances that without experience would probably sound like a random clutter of notes. Recently I've seen two papers that might relate to your question: Jarno Seppanen, "Tatum Grid Analysis of Musical Signals", Proc. IEEE Workshop on Applications of Signal Processing to Audio and Acoustics 2001, pp. 131-134. Jean Laroche, "Estiating Tempo, Swing and Beat Locations in Audio Recordings", Proc. IEEE Workshop on Applications of Signal Processing to Audio and Acoustics 2001, pp. 135-138.M Jim Beauchamp University of Illinois at Urbana-Champaign j-beauch(at)uiuc.edu http://www.staff.uiuc.edu/~j-beauch ------------------------------ Date: Sun, 7 Apr 2002 23:01:17 -0400 From: mercuri(at)GRADIENT.CIS.UPENN.EDU Subject: Re: maximum 'tatum' speed perception As an amateur radio operator, it comes to mind that there are likely a number of military studies on speed of CW ("Morse Code") deciphering. Just as a rough guestimate, I happen to know a number of Extra Class Hams who can receive in the 50 wpm range. Given that a word might average say 5 characters in length, and a character averages 3-4 bits (dits or dahs), that would yield about 750-1000 bpm -- interestingly very similar to the 800 bpm the drummer noted he could hear. All hams find that they have to develop the ability to hear words "as words" and not as letters, in order to pass the 20 wpm test (there's a mental hurdle at 10 wpm too, where letters have to be heard "as letters" instead of dits and dahs). That would be very similar to hearing patterns and phrases in music. I though you folks might be interested in my "back of the envelope" calculations, but I suspect there might be real data about this somewhere (though it may still be classified, who knows). Dr. Rebecca Mercuri KA3IAX (General Class). ------------------------------ End of AUDITORY Digest - 5 Apr 2002 to 7 Apr 2002 (#2002-50) ************************************************************