The Puzzling Mel Scale

The mel scale

http://en.wikipedia.org/wiki/Mel_scale

has been annoying musicians since 1937. Why?

Because it implies some musically puzzling claims, for example that a musical twelth from 233.33 Hz to 700 Hz should somehow sound like being the same size as an octave from 700 Hz to 1400 Hz or a harmonic seventh (4:7) from 1400 Hz to 2450 Hz. To me the twelth and the octave have some similarity but there's no way I'm going to hear the harmonic seventh as being the same size as either of those. Then again, I suppose the scientists didn't play musical intervals to the test subjects. The test subjects were also "musically naive", whatever that means.

The mel scale has been much criticized but it is still finds a lot of use in speech research and there
seems to be a correlation with tonotopic distance in the cochlea. So maybe there is something to it after all.

I was wondering if there is some other property of loudness-matched pure tones that the musically naive test subjects found more salient than their pitch. Loudness is out of the question, so it would have to be some dimension of timbre. Now the timbre of pure tones does seem to range from dull to bright to piercing as frequency is increased. So maybe the mel scale measures brightness rather than pitch. Brightness and pitch are also not independent in pure tones so that would explain the confusion.

Kalle Aho

Hi Kalle!

This may be intersting.
But too bad there isn't a soundfile on wiki.
Have you tried it yourself?

Seems to me though that even if it would work for me it'd only work for a
chromatic scale going upwards?
And defenately not for harmony.

Marcel

The mel scale
>
> http://en.wikipedia.org/wiki/Mel_scale
>
> has been annoying musicians since 1937. Why?
>
> Because it implies some musically puzzling claims, for example that a
> musical twelth from 233.33 Hz to 700 Hz should somehow sound like being the
> same size as an octave from 700 Hz to 1400 Hz or a harmonic seventh (4:7)
> from 1400 Hz to 2450 Hz. To me the twelth and the octave have some
> similarity but there's no way I'm going to hear the harmonic seventh as
> being the same size as either of those. Then again, I suppose the scientists
> didn't play musical intervals to the test subjects. The test subjects were
> also "musically naive", whatever that means.
>
> The mel scale has been much criticized but it is still finds a lot of use
> in speech research and there
> seems to be a correlation with tonotopic distance in the cochlea. So maybe
> there is something to it after all.
>
> I was wondering if there is some other property of loudness-matched pure
> tones that the musically naive test subjects found more salient than their
> pitch. Loudness is out of the question, so it would have to be some
> dimension of timbre. Now the timbre of pure tones does seem to range from
> dull to bright to piercing as frequency is increased. So maybe the mel scale
> measures brightness rather than pitch. Brightness and pitch are also not
> independent in pure tones so that would explain the confusion.
>
> Kalle Aho
>

--- In tuning@yahoogroups.com, Marcel de Velde <m.develde@...> wrote:
>
> Hi Kalle!
>
> This may be intersting.
> But too bad there isn't a soundfile on wiki.
> Have you tried it yourself?

Hi Marcel,

yes, I listened to various melodic pure tone intervals which I created
with the program Goldwave. I compared intervals which are different in
frequency ratio but have the same distance in mels.

> Seems to me though that even if it would work for me it'd only work for a
> chromatic scale going upwards?

The test subjects had to divide pitch differences into parts
that appeared equal to them. I read somewhere that it was later
shown that the results differ depending on whether the tones are
presented in ascending or descending order. This phenomenon is
called "hysteresis".

> And defenately not for harmony.

I didn't even try simultaneous pure tones.

Kalle Aho