[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: seeking Klemm papers



On the issue of the history of the ITD cue for binaural localization, some progress. Stefan Strahl copied Otto Klemm's 1918 and 1920 papers, and sends the report appended below.

I've found several somewhat skeptical contemporaneous reports on Klemm, which I can understand in light of the fact that Klemm reports a few datapoints that are unbelievable outliers. He had one subject that could detect 10 microsecond ITD and "occassionally down to 2 microseconds". The next best was 20 microsecond, which is more believable.

There are also several earlier studies of ITD, or approaches to the concept of ITD besides just interaural phase of pure tones, that predate Klemm 1920, starting with this nice observation on the shock wave of bullets by Mallock in 1908:
http://books.google.com/books?pg=PA110&dq=mallock+%22sensibility+of+the+ear%22&ei=3RCQTd_ZD4SksQPSwuH_CA&ct=result&id=KM0BAAAAMAAJ
which I have read was confirmed by Aggazzotti in 1911 (I should have a copy of his Italian paper by later today probably).

Also, earlier in 1920 and referenced by Klemm 1920 is the paper by Hornbostel and Wertheimer, which I have a translation of from "Forty Germinal Papers in Human Hearing" 1969. The editor says it "is sometimes given priority for seriously proposing interaural \delta t as the dominant cue." He reports success at under 30 microsecond time difference "under favorable conditions".

These have all been found and commented on before, which is why I was able to find them: http://www.google.com/search?tbo=p&tbm=bks&q=Aggazzotti+mallock

Even after all these, it took a while to displace the "phase" idea with the "time" idea, largely due to the Helmholtz influence, I think. For example, Hartley and Fry investigated the localization of complex tones in 1922,
http://books.google.com/books?id=eaW5AAAAIAAJ&pg=RA1-PA33
but they really missed the opportunity to exploit transients when they said:
"It has virtually been assumed that the latter involves three processes: first, the resolution of the sound into its component tones; second, the independent (generally subsconscious) location of each separate component; and third, the formation of a conscious judgment of the position of the source based on the locations of the individual images."

Write me if you want copies of any of this.

Dick

---- Stefan said I can quote him: ----

Please find attached the 1918 and 1920 papers of Otto Klemm. If you need help in translating them please let me know!

I just quickly browsed through them - I hope that I will find some time to read them in detail the next days.. :)

1918, p.88 (end of page)
He had a movable sound source S and in 1 m distance two fixed microphones M_1 and M_2 which were 0.7 m apart. In another room the subject listened to M_1 and M_2 via a kind of headphone with the speakers T_1 and T_2. The stimulus was a "some second long" tone (n=384). He moved the sound source S parallel to the imaginary connection line between M_1 and M_2. The threshold was 6.9 cm which corresponds to 4 degrees. In a second experiment he turned only T_1 or T_2 on (monaural condition) and measured a detection threshold of 10.5 cm.

1920, p. 117
This part is very read-worthy, he gets very emotional :) I'll did a quick translation: He states that in the former experiments he used a "Wundtsche Spaltpegel" which had a time resolution of about 1 \sigma (=1e-3 seconds). He continues that despite the fact that some localisation tasks were observed above 1 \sigma, still a lot was no examined for shorter time intervals so that a repetition of the experiments with finer tools seemed essential. "What a surprise did this simple experiment gave! I assumed that at the best a coarse fraction of 1 \sigma would reveal itself in the perception: but in reality it were first the tenth, then the hundredth of 1 \sigma, and despite all assumption in some single cases the time difference was even smaller. I hoped each day to find the error that would explain such an unlikely accuracy of sensibility: but instead it was confirmed again and again, also from subjects that took part for a casual verification of the experiments. It was like a view in a world of dwarf-like scales. ..."

1920, p. 125
He defines 1 \sigma as 1/1000 seconds and 1 \sigma\sigma as 1/1000000 seconds and reports a localisation threshold of 10 \sigma\sigma, thus 0,01 \sigma (10e-6 seconds), see Table 2. The first column says "subject" / "binaural time threshold" / "occurrence of a simple subjective listening field" / "localisation in the context of time differences" / "illusiveness movement in the context of time differences" / "number of trails". Subject "Kirschmann (A)" showed the best localisation threshold of 0,002 \sigma (2e-6 seconds).

---
Strahl Stefan, Dr. rer. nat. M.Sc. Dipl.-Math.
Research & Development Engineer

MED-EL
Worldwide Headquarters
Fuerstenweg 77a
6020 Innsbruck, Austria