Data from deTune Battle


deTune Battle banner
The most popular page on this blog is a description of an experiment I ran exploring the detuning of music to 432 Hz. The re-pitching of music has a cult following on YouTube. Someone asked me for the raw data from the experiment, and so I thought I might as well publish it here so anyone can download it. If you use the data, let me know by adding a comment below.
If you take the experiment, then it will probably be easier to understand the databases.
The sound files used in the experiment can be downloaded from here (50 MB). If you rebroadcast them, you need to credit the artists listed on this page.
The subject responses for the 1,396 ratings I analysed in december 2013 can be downloaded here. Now, 16,804 responses have been given to the experiment, and the full set can be downloaded here. (Many of these were gathered after I published my blog, so knowledge of the experimental method was known and so might bias the ratings.)
The database columns are described below:

id

A simple counter of the number of responses

userid

Each participant was given a unique ID

age

Respondent age was gathered in decades: 0 corresponds to 0-9 years old, 1 to 10-19 year old, etc. 8 corresponds to 80 and over.

Reproduction

Answers to the question, “I am listening to this experiment using?”
0: headphones
1: laptop/tablet/mobile internal loudspeakers
2: external loudspeakers
3: other or don’t know

Location

Answers to the question, “The place where I am doing this experiment is?”
0: very quiet
1: quiet
2: noisy
3: very noisy

Screennumber

How many sounds each person has heard minus 1. (The counter starts from zero)

SoundA

The first sound heard in the pair

freqA

The tuning for sound A. The music was assumed to be at 440 Hz, all other tunings were achieved using professional pitch shifting software.
For example is SoundA was 15 and FreqA 416, then the sound file heard was 15_416.mp3 (or 15_416.ogg)

SoundB

Second sound in a pair

freqB

Frequency of sound B

Score

The comparative rating of SoundA and SoundB on a 7 point scale, where -3 is “I much prefer A” and 3 is “I much prefer B”.

browser

Browser being used.

Follow me

0 responses to “Data from deTune Battle”

  1. Trevor: Cool you took this on. I don’t have an opinion one way or another as I have experience structuring tunings for different results in many ways. However, I will add to the conversation that detuning previously recorded music that had a foundation of A440 to another starting pitch of any sort is not effectively testing anything significant. This is just not what the issue is all about at all. To test apples to apples one must create music structurally based on A432 and then compare to the same music based on A440 to have a chance at any responses or conclusions that would mean something.

    • I simply copied what people on Youtube have been doing where pitch shifters have been used to retune recordings. You can probably guess my feelings on whether this is a good thing to do or not.

  2. It’s about the same internal coherence of a piece of music being compared to another that has as a starting point different pitches for A. That would cool if you took that one on for folks.

    • Oh I understand Trevor. You were repeating things that people were doing to show how ineffective it is. An actual study to show something significant would be very cool as it would get at the issue and provide insight for people.
      Would not take any more effort than the research you already did to set up and run the experiment. If you want to do it let me know. I have a nuance to add in that could possibly show something else interesting.

  3. The ‘perfect pitch’ idea is interesting. Our Edwardian piano has drifted downwards by almost a semitone over its 100 years, and it has occurred to me that part of the reason why I like its sound could be that it is as though I am transposing everything I play to a different key, or absolute pitch – simply that it is more ‘interesting’ that way because it is different. i.e. that maybe we all, subconsciously, have some element of perfect pitch ability. The pitch-shifter software idea is a perfectly valid way of testing this, I would say.
    However, there are several aspects that would need separating. For example, simple pitch shifters can simply shift the entire spectrum (giving rise to ‘Pinky and Perky’ when shifting upwards) or, on a single source, can be cleverer and attempt to separate the formant and excitation responses. People may like the Pinky and Perky sound (or its opposite when shifting downwards) in itself, or may like the other audible artefacts of different algorithms.
    (There is an actual example of this: how many people bought Babylon Zoo’s song Spaceman for the few seconds of Pinky and Perky at the start, alone?
    https://www.youtube.com/watch?v=GeUx5sdp3qM )
    Creating acoustic music from scratch at a different ‘A’ frequency would not be unambiguous, either. People may be responding to the change of absolute pitch or may be responding to novel resonances and subtle changes in instruments that were ‘designed’ for A440, or to different resonances in the room, or they may be responding to what they perceive as different ‘orchestration’. Separating the various effects would be an interesting problem to solve.

  4. Hi Trevor,
    I am a college freshman taking a statistics course. I would love to use your raw data for my personal statistics project, and I will most definitely give you the credit for the data. Thank you for putting this amazing data up for use!

Leave a Reply to Alan Tower Cancel reply

Your email address will not be published. Required fields are marked *

%d bloggers like this: