Re: Re: Re: Re: Re: Re: Re: Re: OK, I did it - update


[ Follow Ups ] [ Post Followup ] [ TubeNet BBS ] [ FAQ ]

Posted by Rick Denney on October 23, 2001 at 11:24:21:

In Reply to: Re: Re: Re: Re: Re: Re: Re: OK, I did it - update posted by js on October 23, 2001 at 02:48:25:

For those of you who might be looking for a student project or some similar activity, a brief discussion of how to do a scientific study might help get you started. I'm not writing this to complain to Joe--he tested what he tested and rendered his opinion. I'm more interested in those who wonder about what it takes to do it scientifically. Here it is:

The opinions of qualified listeners can indeed be scientific, if a sound difference is what you are testing. If you say something sounds different (or doesn't), then a listening test is the only way to answer it. But you have to remove the bias of the player, and the bias of the listeners. Here's how I would do it:

Disguise the mouthpieces. Take a heavy mouthpiece and a light mouthpiece of identical interior shape, and encase both in a large styrofoam block to completely cover up their exterior shape. Both mouthpieces must be visually identical. Have them brought to the tuba and mounted by a researcher (not by the player), so that only the researcher knows which is which, following a strict protocol of saying nothing to the player. Have the player play a short set of music, behind a screen, as at an audition. Use a DEG Tuba Rest if the weight of the mouthpiece differs enough to be noticeable even when already mounted on the instrument.

The listener's panel would comprise perhaps half a dozen qualified listeners, but not necessarily tuba players. They would be given a list of characteristics to rate for each playing set, including all those that seem of interest to the researchers, and a few that don't. An example: "Rate the quality of the sound in terms of projection: 1 (sound is dim and unfocused) 2 3 4 5 (sound is piercing and direct)." You can also ask for general comments about each trial.

Play the set ten times, five times with one mouthpiece and five times with the other. Before each set, the researcher brings out a mouthpiece, installs it on the tuba, and leaves. The player plays the set. The research uses a randomized series to determine which set will get which mouthpiece, with no indication given to either the tuba player or to the listeners.

Then you analyze the results. Divide the listener's answers into two groups--those with the heavy mouthpiece and those with the light mouthpiece. If the differences show any significant differences (as statistically analyzed), then you have your difference. If they don't, then you can say that weight made no noticeable difference to the listeners. If the general comments show a difference but the numbers don't, then you may be asking the wrong questions, or you may need more trials for a difference to reveal itself.

You can also have the player answer a short questionnaire about playability issues after each set.

This satisfies all your requirements, because the inside shape of the mouthpiece is identical in both cases, and because that shape can be the player's favorite shape. If you want to broaden the study, you can make up mouthpiece pairs to suit the peferences of other players on different instruments. It also satisfies science, because it is a double-blind approach (both the person administering the test--the player--and the persons evaluating the test--the listeners--are unaware of which mouthpiece is being used). The biases of both are therefore filtered out. And it also tests just what the mouthpiece is supposed to provide: a different sound or a different feel.

You can also measure the sound directly using a spectrum analyzer, but that only studies one aspect of the sound, and it doesn't directly measure the effects that are interesting to tuba players. Who cares if the 12th partial is made louder (or softer) by the special mouthpiece, if we don't know whether it is good to have a stronger (or weaker) 12th partial? You can even model the physical effects analytically, and produce a theory based on that analysis that can subsequently be tested. That may give you what you need to determine when it is better and on what sorts of tubas. But the empirical approach with a panel of listeners will get you a relevant answer sooner.

Rick "offering some science" Denney


Follow Ups: