OT Hearing Receptors

[bpodruzny]

I’m on a study break from my Neuroscience and I thought I’d share my speaker related revelation of the day.

So, according to this textbook, audio receptors are most sensitive at 2000hz and have a sharp fall off from there +/- . If I interpret the chart correctly 1000hz to 5000hz is still quite sensitive. From an evolutionary standpoint this makes a lot of sense, 2000hz is right in the middle of vocal spectrum. What I find fascinating from a speaker-building standpoint is 1000 – 5000hz (and a bit higher 7000hz IIRC) is what other post on PE have extolled as the “most crucial range to have a flat response”.

This makes me wonder….

What if you used extended range driver to cover from 300-5000hz range, a good 10” for the bottom end, and the Dayton Neo to fill the highs. Just a thought, but please let me know what you think!

Cheers
-Brian

[davidvwebber]

> I’m on a study break from my Neuroscience
> and I thought I’d share my speaker related
> revelation of the day.

> So, according to this textbook, audio
> receptors are most sensitive at 2000hz and
> have a sharp fall off from there +/- . If I
> interpret the chart correctly 1000hz to
> 5000hz is still quite sensitive. From an
> evolutionary standpoint this makes a lot of
> sense, 2000hz is right in the middle of
> vocal spectrum. What I find fascinating from
> a speaker-building standpoint is 1000 –
> 5000hz (and a bit higher 7000hz IIRC) is
> what other post on PE have extolled as the
> “most crucial range to have a flat
> response”.

> This makes me wonder….

> What if you used extended range driver to
> cover from 300-5000hz range, a good 10” for
> the bottom end, and the Dayton Neo to fill
> the highs. Just a thought, but please let me
> know what you think!

> Cheers
> -Brian
That is what my JBL " Baby Summits " do !
A 14" Woofer up to the 300 / 500 point ,
a LE85 Mid up to where the Tweter takes over !!

[damkor]

...audio
> receptors are most sensitive at 2000hz and
> have a sharp fall off from there +/- ...From an
> evolutionary standpoint this makes a lot of
> sense, 2000hz is right in the middle of
> vocal spectrum.

2000Hz is a bit high. Mariah Carey has sung the highest note ever, and I think that was barely 2,000Hz. Somewhere between 100 and 1000Hz is more like it, depending on ***, age etc.

Anyway, an interesting question is whether we have evolved to listen to each other speak or hear screams, predators and prey, rustling. A bit of both I think.

[Dan B]

I just opened Win Isd played a 1,000hz tone and I could match it in falsetto with my voice (I'm a bass)

[bpodruzny]

Sorry about the voice thing, 2000hz would be a very high range. It likely is more for hearing screams of warning/ hearing our children. What I found interesting is the most sensitive zone is right here most crossovers cross in 2 ways. Perhaps this is a physiological support for the use of 3-ways?

Anyway, off to write the neuroscience exam, it’s a good thing I don’t have to know what frequencies correlate with specific sounds.

Cheers
-Brian

[damkor]

> I just opened Win Isd played a 1,000hz tone
> and I could match it in falsetto with my
> voice (I'm a bass)

Sure, but people, even small children, don't usually speak using tones any higher than that. 1000Hz is almost two octaves above middle C, which is 262 Hz. Actually closer to the B a semitone below it. Are you sure you weren't singing an octave lower? It's hard to tell sometimes.

<A HREF="http://en.wikipedia.org/wiki/Piano_key_frequencies">http://en.wikipedia.org/wiki/Piano_key_frequencies</A>

[damkor]

The highest note ever recorded sung by a human voice. (Although there is controversy over whether she was a little flat, maybe it was only 3150Hz! And there is always the potential for electronic manipulation nowadays.) She's also the most successfull female composer in history.

[wg_ski]

> Sorry about the voice thing, 2000hz would be
> a very high range. It likely is more for
> hearing screams of warning/ hearing our
> children. What I found interesting is the
> most sensitive zone is right here most
> crossovers cross in 2 ways. Perhaps this is
> a physiological support for the use of
> 3-ways?

> Anyway, off to write the neuroscience exam,
> it’s a good thing I don’t have to know what
> frequencies correlate with specific sounds.

> Cheers
> -Brian

That's right - the fundamentals of voice and most instruments are WAY below 2kHz. But it is the harmonic structure (as well as dynamics) that give each instrument its timbre - and that's how we can tell them apart. That information is centered higher in frequency - and up to a couple kHz is very significant. Response anaomalies in the 2k region are thus much more evident than something around, say, 300 Hz. As you go high, again the energy falls off, and most of what is up there is percussive sounds which are more forgiving due to their wide-band energy content. Cymbals and such produce more of a noise (band) spectrum than any particular note.

The old school way of building speakers took full advantage of this. Using a wide band midrange that's as flat as possible and a dedicated woofer and tweeter to get to what the midrange couldn't can work well even if the crossover isn't perfect. At "tweeter" crossover frequencies above 4k or so, phase mistracking can easily be buried in with the effects of diffraction and energy storage and no one would be the wiser. At "woofer" crossover frequencies, both drivers tend to remain pistonic and have little or no excess phase so a textbook filter response actually gives the desired result.

[dickwest]

IIRC Bell Labs, back in the 1930s, concluded that for telephone speech intelligibility only 300 to 3000 hz was required.

Dick

[damkor]

> That's right - the fundamentals of voice and
> most instruments are WAY below 2kHz.

Sopranos frequenctly hit above 2kHz. That is, the fundamental tone they are producing, the one that lights up with the most amplitude on a frequency analyser, is above 2kHz.

It is tricky to tell though if you are not a prof. singer. Try playing a note on a piano, and singing it. You can usually sing an octave below or above and have it sound darn like the note on the keyboard, by virtue of the harmonics that you are making, that give the illusion of a higher or lower note.

[damkor]

...an extra level of visceral realism to heavy breathing over the telephone, for example. I think they were right. Still, speech over the telephone may be highly intelligible, but it certainly doesn't sound anything like a live voice.

[Max_Andrews]

Provided Link: equal loudness contours


We are actually most sensitive to the 5000hz range.
What is perhaps more interesting however is the degree of the difference. A 2000hz tone playing at 65dB and a 40Hz tone playing at 90db --about 500 times as loud, will sound the same volume to our ears. This is why woofers need to be a lot larger than tweeters and have a lot more travel. We need the bass and midrange to play far louder in order to match our hypersensitivity to the high frequencies.
What is also interesting is that as the volume of the sound increases, our perception becomes more linear. At 90db the human ear has the flattest frequency response between 20hz and 3000hz, and this is often why music sounds best at around this volume.

[dickwest]

> We are actually most sensitive to the 5000hz
> range.
> What is perhaps more interesting however is
> the degree of the difference. A 2000hz tone
> playing at 65dB and a 40Hz tone playing at
> 90db --about 500 times as loud, will sound
> the same volume to our ears. This is why
> woofers need to be a lot larger than
> tweeters and have a lot more travel. We need
> the bass and midrange to play far louder in
> order to match our hypersensitivity to the
> high frequencies.

Ah . . . Yes. Curves of equally perceived loudness (Phons). I once met Dr. Fletcher when I was much younger. He tried to talk me into an engineering major, so I chose psychology (duh!).
> What is also interesting is that as the
> volume of the sound increases, our
> perception becomes more linear. At 90db the
> human ear has the flattest frequency
> response between 20hz and 3000hz, and this
> is often why music sounds best at around
> this volume.

[rone]

The wider the range you cover, the more IM distortion becomes a factor.

People make a big deal about crossing over in the vocal range - I think you are more likely to hear a 5k crossover because that is too high for all but a very small driver, which would not make it to the low range you want without being very directive at high frequencies. Smooth power response is as important as flat frequency response.

[kene]

> The highest note ever recorded sung by a
> human voice. (Although there is controversy
> over whether she was a little flat, maybe it
> was only 3150Hz! And there is always the
> potential for electronic manipulation
> nowadays.) She's also the most successfull
> female composer in history.

That sounds closer to what I've heard her pull off in some of her songs. I'm not a perfect pitch person or anything, but I've heard 2000 Hz enough times that it didn't quite sound right.

[kene]

> IIRC Bell Labs, back in the 1930s, concluded
> that for telephone speech intelligibility
> only 300 to 3000 hz was required.

> Dick

Probably a +/- 3dB figure, though, with a rolloff that continues past that range. When I was a kid I was shown the wave forms of numerous spoken sounds on an oscilloscope. The one that was most striking to me (which probably has something to do with why it's the only one I remember) was "M" versus "N". They looked virtually identical, I think with a subtle square-wave component. The only visible difference was a very small bump which showed up either at the beginning or the end of the square. Even as a wee kid it was amazing to me to think that our ears could resolve such a seemingly minute detail with such ease, and I'm only more amazed now when I think about how we additionally do it with enormous amounts of overlaying noise often within the same frequency range (other instruments in music, background noise in the office, and so on). It also stands to me as good evidence for why the limited set of measurements we usually talk about in audio components is horribly inadequate for discribing sound quality, but I digress. Anyway, my original point is that if that small bump is all that cues us off between "N" versus "M" and it's probably hovering around ten times the fundamental frequency, 300-3,000 would be inadequate for speech resolution if it were an absolute cutoff versus a gradual roll-off.

[skebenin]

> The one that was most striking to me (which
> probably has something to do with why it's
> the only one I remember) was "M"
> versus "N". They looked virtually
> identical, I think with a subtle square-wave
> component.

The difference would probably be more pronounced if you looked at it in fourier plot. The cochlea is basically a fourier device.

[kene]

> The difference would probably be more
> pronounced if you looked at it in fourier
> plot. The cochlea is basically a fourier
> device.

True that, but amazing nonetheless.

[damkor]

I can't stand her style myself. That Miasma cr*p?! It's musically quite dull, except on some sort of technical level maybe and doesn't have a drip of soul. I can appreciate her technical skill though, and her success in turning that into mega-success.

[kene]

> I can't stand her style myself.

Yeah, I've never felt particularly impassioned by her work, but I haven't listened to much. What little I've heard seems to flaunt more of that wide bandwidth than instill passion or soul.

But that's alright, I'm sure she's a nice person and that's really all that matters and in no way am I saying that because she also happens to be utterly hot.
:-b