HT wine-tasting (was HT tests)

[Jason Kanter]

All right, I admit, there are in fact two recordings as described in the initial post of this thread. Same pianist, same program, same model piano, same recording equipment. Different hall, different tuner, different temperament. (No Disklavier versions.) (And I haven't yet heard them.)

I propose creating two CDs, similar in construction but with a different sequence of temperaments, and distributing them so that the recipients do not know which version they have received. Each CD to be identified by a serial number. Each CD will have a series of pairs of tracks, e.g. track 1A and 1B, 2A and 2B, 3A and 3B etc. For each pair of tracks, the musical segment will be the same, but A and B are taken from different performances -- one ET, the other HT. The "A" and "B" are randomized -- sometimes the A is ET, sometimes the B is ET. And although the sequence of musical segments is the same on both versions of the CD, the distribution of ET and WT into the A and B slots is different -- so no one except the distributor of the CDs knows which is which. This satisfies the need for the "double-blind" testing criteria. The host of the HT wine-tasting will not be able to reveal his/her personal knowledge through facial expressions or body language.

At the wine-tastings, people listen to pairs of tracks, and during/after each pair they fill out an eval form that asks two kinds of questions: (1) related to how musically pleasing the segment was, and (2) related to which temperament was used in that segment.

(1) For the musical preference test, the eval form asks the listener to rate:
1. It's quite musically pleasing, better than the other track.
2. It's okay, and so is the other track.
3. There's something musically displeasing about it, or it doesn't sound as good as the other track.

(2) For the temperament identification, the eval form asks the listener to check one of these:
1. I'm sure it's ET.
2. I guess it's ET.
3. I'm not sure.
4. I guess it's HT.
5. I'm sure it's HT.

Now my question turns to statistical validation. How many CDs should be issued? (How many people would volunteer to host such a wine-tasting? A very interesting evening it would be!) How many listeners should be polled? (The more the better, but how many would it take to satisfty the skepticism of e.g. ric moody?) Are the above questions and rating scales the right ones? Are there other questions that should be asked? Should the two tracks (1A and 1B) be played just once, or repeated, or only repeated if listeners so request?

Note: I am suggesting ONE CD with alternating tracks from both performances for the reason that this will better enable direct comparisons. The musical program that has been recorded is about 75 minutes long, I believe. That's too long for this kind of comparison, in my opinion. Better to use relatively short "movements" side-by-side. 

Whaddyathink?
..........
.  jason kanter . jason.kanter@wamu.net . jkanter@rollingball.com
.  vp/manager . learning & performance development
.  consumer loan servicing . washington mutual bank
.  office 206 490 6708 . cell 425 830 1561
...............................

----- Original Message ----- 
From: "Richard Moody" <remoody@midstatesd.net>
To: "Pianotech" <pianotech@ptg.org>
Sent: Wednesday, March 19, 2003 7:45 PM
Subject: Re: HT wine-tasting (was HT tests)


| 
| ----- Original Message -----
| From: <A440A@aol.com>
| To: <pianotech@ptg.org>
| Sent: Tuesday, March 18, 2003 3:03 PM
| Subject: Re: HT wine-tasting (was HT tests)
| 
| 
| >At this stage of the game,
| > establishing the preferences of the "listening group" is a
| valuable >piece of information.
| >     As to "why" these preferences are observable, I submit
| .......
| 
| Unfortunately the "preferences" gleaned by a test, of "which
| sounds better" sample A, or Sample B is not much information at
| all.  The subject has only two choices and knows they are expected
| to pick the "best" answer.  Now if a third choice is included,
| "Can't really tell"  I think you will get a better understanding
| of how much HT's or OT's  matter to the "qualified listener" and
| such blind tests might shed a whole new light on the debate,
| possibly favorable to the hypothesis, "it doesn't really matter".
| 
|     My test is to have the practitioners of HT's listen to CD's
| and tell which are ET and which are not.   After all if HT's make
| such a difference shouldn't they be able to tell?  So far no
| takers.
| 
| How  bout this test.... ?
| 
| Listen to these samples of  pianos tuned in  HT, or ET.   After
| each selection please mark the following......
|     A. ET
|     B.  HT
|     C.  Can't tell.
|     D. Care to guess?
|             1. HT
|              2. ET
| 
|     -----rm
| 
| 
| 
| _______________________________________________
| pianotech list info: https://www.moypiano.com/resources/#archives