Reply by Bob Masta February 9, 20152015-02-09
On Fri, 6 Feb 2015 17:42:44 +0000 (UTC), glen herrmannsfeldt
<gah@ugcs.caltech.edu> wrote:

>>>Bob Masta <N0Spam@daqarta.com> wrote:
<snip>
>> Consider a fundamental of (say) 110 Hz, which excites a >> neural pathway at a firing rate proportional to the >> amplitude of the fundamental. To activate that same >> pathway when the fundamental is missing and only harmonics >> are present, we need to detect the presence of activity on >> the 220 Hz line AND the 330 Hz line, and various >> combinations of higher harmonic lines. Those lines are >> right there, we just need a few "analog AND gates" to >> combine them. > >At some point, there is a system that does pattern recognition. >Is this a sound that I have heard before? But the sounds won't >have exactly the same spectral content, so it has to be able >to match them even a little different. That probably works better >for more natural sounds, such as human voice, and less for sums >of a small number of harmonics, though. Seems that it wouldn't >be hard to test different combinations of harmonics at different >amplitudes to see though.
>> A single neuron is perfect for this; that's what it normally >> does. It sums/integrates all input synapse contributions, >> and if-and-when the sum reaches threshold, it fires. To get >> an AND effect, the proportions of input synapses ony need to >> be such that the (near) simultaneous contributions from 2 or >> more harmonic lines are needed to reach threshold. (An OR >> is when each input line has enough moxie to reach threshold >> on its own.) The integration period (volume of ions that >> need to be pumped out of the cell to reach threshold) >> controls how "simultaneous" the harmonics need to be. > >It is supposed to be that babies learn pretty young which >tones are important in the spoken languages that they hear. >Past some age, maybe about one, those patterns are pretty >much fixed. Since human voice has pretty many harmonics, >that would go into the tone detection.
YES! You have added the piece I was missing... it's all down to pattern recognition! I suspect missing fundamental detection could easily be demonstrated with a simple "neural network" approach. There would be an input line for each frequency from the cochlea, and an output line for each frequency to the brain. When we first start hearing sounds (in the womb, even) the network would reinforce pathways that typically were activated together, such as harmonics present along with the fundamental. Then it would recognize this pattern even if some components were missing, possibly including the fundamental. This sort of pattern recognition doesn't require any delays that normal autocorrelation methods seem to require, nor even anything about phase, or the raw waveform. The brain has everything it needs to do this with only the one-fiber-per-frequency we already know exists. It's still performing a "correlation", but it has a head start by already having the frequencies separated. My AND/OR scheme might, or might not, be somewhere at the bottom of the neural net nitty-gritty, but the pattern recognition is really the big picture. Many thanks, and best regards, Bob Masta DAQARTA v7.60 Data AcQuisition And Real-Time Analysis www.daqarta.com Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE Signal Generator, DaqMusiq generator Science with your sound card!
Reply by Bob Masta February 7, 20152015-02-07
On Fri, 6 Feb 2015 17:42:44 +0000 (UTC), glen herrmannsfeldt
<gah@ugcs.caltech.edu> wrote:

<snip>
>Well, as I understand it the Q is pretty low, at least by DSP >standards, so many nearby frequencies will also have a signal. >
I've also heard people suggest that the Q is low, but I'm not sure where that comes from. When you record from the IHCs you find the Q is *amazingly* high, with slopes of 100s of dB/octave. For IHCs, Q is (or was when I was involved) reported as "Q 10 dB" instead of the normal 3 dB type. That goes back to the original papers by Russell and Sellick (1978) that were the first to report intracellular recordings. They reported values in the 5-8 range, which may be what people are referring to. However, the high Q is only if the *outer* hair cells are functioning properly, and the cochlea is in good physiological shape by the time you get your electrode into the IHC... *very* difficult to achieve after surgery to expose the cochlea. Note that the OHCs act as electromechanical amplifiers to supply positive feedback that sharpens the tuning. But this effect is damped down by the brain via efferent fibers *to* the OHCs which are activated in the presence of loud sound. (The OHCs don't send afferent fibers to the brain... that's done strictly by IHCs.) So I wonder if the low-Q comments are based on non-physiological cochleas, which is what you'd get about 9 guinea pigs out of 10. (Note that all the pioneering Bekesy work was done on *dead* cochleas, so of course they had low Q.) Alternatively, the Q of an IHC goes down as sound level goes up, so the low-Q reports might have been from loud-sound studies.
>At some point, it has to be processed such that we know it is >a single tone.
In the visual system there is a mechanism called "lateral inhibition" that sharpens contrast, whereby cells that are stimulated (by light) supply inhibitory signals to their neighbors. Back in the Early Days of hearing research, this seemed like an obvious explanation for the sharp frequency resolution of the ear. But (at least when I was involved) nobody had found anything like this in the auditory system. Yet it seems like too good a deal for the brain to have overlooked, so it might be used at higher levels to sharpen tuning for loud sounds. Since I've been out of the game, people have made a lot of progress recording from the cochlear nucleus and higher, so they may have found lateral inhibition by now. At any rate, wherever the single-tone determination is achieved, it seems to me that a simple circuit of neural ANDs and ORs can add in Missing Fundamental information. Best regards, Bob Masta DAQARTA v7.60 Data AcQuisition And Real-Time Analysis www.daqarta.com Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE Signal Generator, DaqMusiq generator Science with your sound card!
Reply by glen herrmannsfeldt February 6, 20152015-02-06
Bob Masta <N0Spam@daqarta.com> wrote:
 
>>Bob Masta <N0Spam@daqarta.com> wrote:
(snip)
>>> There are problems with envelope detection, and especially >>> hair cell rectification, as explanations for the missing >>> fundamental. Consider that the cochlea behaves like a bank >>> of mechanical filters, essentially one for each frequency we >>> can resolve. The (inner) hair cell is the detector *after* >>> a filter, so electrical rectification there does not >>> introduce any mechanical distortion product that can >>> propagate to the filter for the missing fundamental >>> frequency.
>>OK, but say you add up the signals from all the hair cells.
>>I don't know if it helps, but as I understand it, told to me >>by someone who does actual experiments, that hair cells send >>out nerve impulses at the peak of the sine for low frequencies, >>and approximately every 2nd or 3rd, or maybe more, for higher >>frequencies.
> True, considering the inner hair cell and the neuron it > feeds as a system. (The IHC doesn't actually generate > impulses, it has a "graded" response in releasing > transmitter molecules onto a very long "dendrite" fiber, > whose cell body is in the cochlear nucleus and is the neuron > that actually generates the spikes.)
Yes. The actual reason I asked this years ago was the age old question of phase. If the signals from the cochlea don't preserve phase, then obviously we can't hear phase, but they do.
> If the system is near the firing threshold, then a peak in > the incoming frequency is more likely to push it over into > firing. Once it fires, a neuron goes into a refractory > period while it recharges (pumps ions across its cell wall > to get back well below threshold) before it can fire again. > That's why its max firing rate is only a few hundred Hz. > But then when it's again able to fire, it will be more > likely on whatever peak arrives next. If you plot a > histogram, you see spikes clumped at periods that match the > tuned frequency, even though many sine peaks may be missed > between sequential pulses.
>>Seems to me that with the right weighting and time averaging, >>one might approximate the envelope.
> Yeah, but then after that weighting and averaging, you'd > only be up to the point of having a putative *input* to a > correlator that might hope to extract the periodicity.
Oh, yes, I was just saying that there could be an envelope detection, not what you would do with one.
> Remember, the brain is looking for one-fiber-per-frequency > inputs. To detect the "missing" fundamental, it needs a > signal on the line that corresponds to that particular > frequency, the same line that would be excited by the > fundamental alone. So your correlator needs a separate > output line for each detectable frequency.
Well, as I understand it the Q is pretty low, at least by DSP standards, so many nearby frequencies will also have a signal. At some point, it has to be processed such that we know it is a single tone.
> But there is a *much* simpler way, which as far as I can > tell nobody is considering (including hearing researchers). > And it doesn't use any tricky circuits like correlators that > nobody knows how to make with ordinary neurons. It's one of > those head-smackingly obvious things, once you look at the > system and its existing components.
> Consider a fundamental of (say) 110 Hz, which excites a > neural pathway at a firing rate proportional to the > amplitude of the fundamental. To activate that same > pathway when the fundamental is missing and only harmonics > are present, we need to detect the presence of activity on > the 220 Hz line AND the 330 Hz line, and various > combinations of higher harmonic lines. Those lines are > right there, we just need a few "analog AND gates" to > combine them.
At some point, there is a system that does pattern recognition. Is this a sound that I have heard before? But the sounds won't have exactly the same spectral content, so it has to be able to match them even a little different. That probably works better for more natural sounds, such as human voice, and less for sums of a small number of harmonics, though. Seems that it wouldn't be hard to test different combinations of harmonics at different amplitudes to see though.
> A single neuron is perfect for this; that's what it normally > does. It sums/integrates all input synapse contributions, > and if-and-when the sum reaches threshold, it fires. To get > an AND effect, the proportions of input synapses ony need to > be such that the (near) simultaneous contributions from 2 or > more harmonic lines are needed to reach threshold. (An OR > is when each input line has enough moxie to reach threshold > on its own.) The integration period (volume of ions that > need to be pumped out of the cell to reach threshold) > controls how "simultaneous" the harmonics need to be.
It is supposed to be that babies learn pretty young which tones are important in the spoken languages that they hear. Past some age, maybe about one, those patterns are pretty much fixed. Since human voice has pretty many harmonics, that would go into the tone detection.
> Now, you'd be right in thinking that this scheme will have > trouble working at different signal levels: After all, if > the spike rates are doubled, a 2-input AND would fire with > only one input active. And if all spike rates are halved, > it might not fire even with both inputs active. I think > this can be compensated fairly easily using an inhibitory > input to act as a sensitivity control.
But it has to work at different levels. We recognize people by their voice at varying levels. Even worse, our spectral sensitivity changes with amplitude, the reason for loudness controls on audio systems.
> Yes, this scheme would take several extra neurons for each > of the few thousand resolvable frequencies. To the brain, > that's chump change. And I suspect it is far cheaper than > any sort of correlation scheme.
I think everyone with a cat (and probably dogs, too) knows how fast they learn the sound of a can opener. (We don't give our cats canned food very often, and they still learn it.)
> This approach seems like "Duh!"-obvious, which might mean > that there is something "Duh!"-stupid wrong with it. So I'm > depending on you guys to swat it down to size!
-- glen
Reply by Bob Masta February 6, 20152015-02-06
On Thu, 5 Feb 2015 14:01:21 +0000 (UTC), glen herrmannsfeldt
<gah@ugcs.caltech.edu> wrote:

>Bob Masta <N0Spam@daqarta.com> wrote: > >(snip) > >> There are problems with envelope detection, and especially >> hair cell rectification, as explanations for the missing >> fundamental. Consider that the cochlea behaves like a bank >> of mechanical filters, essentially one for each frequency we >> can resolve. The (inner) hair cell is the detector *after* >> a filter, so electrical rectification there does not >> introduce any mechanical distortion product that can >> propagate to the filter for the missing fundamental >> frequency. > >OK, but say you add up the signals from all the hair cells. > >I don't know if it helps, but as I understand it, told to me >by someone who does actual experiments, that hair cells send >out nerve impulses at the peak of the sine for low frequencies, >and approximately every 2nd or 3rd, or maybe more, for higher >frequencies.
True, considering the inner hair cell and the neuron it feeds as a system. (The IHC doesn't actually generate impulses, it has a "graded" response in releasing transmitter molecules onto a very long "dendrite" fiber, whose cell body is in the cochlear nucleus and is the neuron that actually generates the spikes.) If the system is near the firing threshold, then a peak in the incoming frequency is more likely to push it over into firing. Once it fires, a neuron goes into a refractory period while it recharges (pumps ions across its cell wall to get back well below threshold) before it can fire again. That's why its max firing rate is only a few hundred Hz. But then when it's again able to fire, it will be more likely on whatever peak arrives next. If you plot a histogram, you see spikes clumped at periods that match the tuned frequency, even though many sine peaks may be missed between sequential pulses.
>Seems to me that with the right weighting and time averaging, >one might approximate the envelope.
Yeah, but then after that weighting and averaging, you'd only be up to the point of having a putative *input* to a correlator that might hope to extract the periodicity. Remember, the brain is looking for one-fiber-per-frequency inputs. To detect the "missing" fundamental, it needs a signal on the line that corresponds to that particular frequency, the same line that would be excited by the fundamental alone. So your correlator needs a separate output line for each detectable frequency. But there is a *much* simpler way, which as far as I can tell nobody is considering (including hearing researchers). And it doesn't use any tricky circuits like correlators that nobody knows how to make with ordinary neurons. It's one of those head-smackingly obvious things, once you look at the system and its existing components. Consider a fundamental of (say) 110 Hz, which excites a neural pathway at a firing rate proportional to the amplitude of the fundamental. To activate that same pathway when the fundamental is missing and only harmonics are present, we need to detect the presence of activity on the 220 Hz line AND the 330 Hz line, and various combinations of higher harmonic lines. Those lines are right there, we just need a few "analog AND gates" to combine them. A single neuron is perfect for this; that's what it normally does. It sums/integrates all input synapse contributions, and if-and-when the sum reaches threshold, it fires. To get an AND effect, the proportions of input synapses ony need to be such that the (near) simultaneous contributions from 2 or more harmonic lines are needed to reach threshold. (An OR is when each input line has enough moxie to reach threshold on its own.) The integration period (volume of ions that need to be pumped out of the cell to reach threshold) controls how "simultaneous" the harmonics need to be. Now, you'd be right in thinking that this scheme will have trouble working at different signal levels: After all, if the spike rates are doubled, a 2-input AND would fire with only one input active. And if all spike rates are halved, it might not fire even with both inputs active. I think this can be compensated fairly easily using an inhibitory input to act as a sensitivity control. Yes, this scheme would take several extra neurons for each of the few thousand resolvable frequencies. To the brain, that's chump change. And I suspect it is far cheaper than any sort of correlation scheme. This approach seems like "Duh!"-obvious, which might mean that there is something "Duh!"-stupid wrong with it. So I'm depending on you guys to swat it down to size! Best regards, Bob Masta DAQARTA v7.60 Data AcQuisition And Real-Time Analysis www.daqarta.com Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE Signal Generator, DaqMusiq generator Science with your sound card!
Reply by glen herrmannsfeldt February 5, 20152015-02-05
Bob Masta <N0Spam@daqarta.com> wrote:

(snip)

> There are problems with envelope detection, and especially > hair cell rectification, as explanations for the missing > fundamental. Consider that the cochlea behaves like a bank > of mechanical filters, essentially one for each frequency we > can resolve. The (inner) hair cell is the detector *after* > a filter, so electrical rectification there does not > introduce any mechanical distortion product that can > propagate to the filter for the missing fundamental > frequency.
OK, but say you add up the signals from all the hair cells. I don't know if it helps, but as I understand it, told to me by someone who does actual experiments, that hair cells send out nerve impulses at the peak of the sine for low frequencies, and approximately every 2nd or 3rd, or maybe more, for higher frequencies. Seems to me that with the right weighting and time averaging, one might approximate the envelope.
> But it's even worse than that: Because of the mechanical > filtering, the hair cell isn't even *exposed* to the > envelope... it just sees the frequency that its tiny patch > of basilar membrane is tuned to. It's like looking at one > line of the output of an FFT of the harmonic series, where > you don't see any envelope at all, just the individual > spectral peaks.
> The envelope is strictly a waveform phenomenon, while the > cochlea is a mechanical spectrum analyzer. There really > isn't any place where the input waveform appears on the > basilar membrane.
(snip) -- glen
Reply by Bob Masta February 5, 20152015-02-05
On Tue, 3 Feb 2015 19:43:27 -0800 (PST),
radams2000@gmail.com wrote:

>It would be interesting to mess around with the phases of the harmonics and= > see if that makes any difference. One theory of the missing fundamental pr= >inciple is that you are sensitive to the modulation of the envelope of the = >signal (remember that the hair cells along the basilar membrane act as half= >-wave rectifiers). The envelope modulation would be affected by relative ha= >rmonic phase.=20
There are problems with envelope detection, and especially hair cell rectification, as explanations for the missing fundamental. Consider that the cochlea behaves like a bank of mechanical filters, essentially one for each frequency we can resolve. The (inner) hair cell is the detector *after* a filter, so electrical rectification there does not introduce any mechanical distortion product that can propagate to the filter for the missing fundamental frequency. But it's even worse than that: Because of the mechanical filtering, the hair cell isn't even *exposed* to the envelope... it just sees the frequency that its tiny patch of basilar membrane is tuned to. It's like looking at one line of the output of an FFT of the harmonic series, where you don't see any envelope at all, just the individual spectral peaks. The envelope is strictly a waveform phenomenon, while the cochlea is a mechanical spectrum analyzer. There really isn't any place where the input waveform appears on the basilar membrane. Curiously, people used to think the cochlea worked on waveforms, since they could put a simple electrode on it and easily record a "cochlear microphonic" gross potential that replicated the input sound wave. (Played back through an amp and speaker, it sounded like a somewhat tinny radio or phonograph, say from the 30s or 40s.) This was considered evidence for the "telephone theory" of hearing, which was later supplanted by Bekesy's traveling wave and "place principle." Best regards, Bob Masta DAQARTA v7.60 Data AcQuisition And Real-Time Analysis www.daqarta.com Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE Signal Generator, DaqMusiq generator Science with your sound card!
Reply by February 4, 20152015-02-04
On Wednesday, February 4, 2015 at 3:03:11 PM UTC-5, Eric Jacobsen wrote:
> On Wed, 4 Feb 2015 09:28:06 -0800 (PST), bellda2005@gmail.com wrote: > > >On Wednesday, February 4, 2015 at 6:54:45 AM UTC-5, Rick Lyons wrote: > >> On Wed, 28 Jan 2015 12:07:52 -0800 (PST), bellda2005@gmail.com wrote: > >>=20 > >> [Snipped by Lyons] > >> > > >> >Hi Rick, > >> > > >> >You have actually run across the "missing fundamental" 1000's of times b= > >efore. Every time you made a phone call on a land line the fundamental at a= > > minimum was largely suppressed if not effectively eliminated (except maybe= > > for squeaky children and extremely high-pitched people), yet you did not t= > >hink the pitch was any different. The period of a waveform rich with harmon= > >ics will be unchanged whether the fundamental is present or not. That perio= > >d is the pitch, is it not? > >> > > >> >An interesting question (not a proof of anything) is if the fundamental = > >is missing for speech, what do you expect the pitch to be? It hasn't double= > >d because the waveform is not periodic in half the period of the fundamenta= > >l. Some other choice? > >> > > >> >In mobile phones that use spectral coding bandwidth compression it would= > > make sense to eliminate low frequencies including the fundamental to save = > >bits, so I would guess it is probably done there, which could be more 1000s= > > of times you have encountered it. > >> > > >> >Dirk > >> > > >> >Dirk Bell, MSEE, MSCF > >> >Senior-Level DSP Engineer in search of a great opportunity. > >> >Willing to relocate. > >> >https://www.linkedin.com/in/dirkbell > >>=20 > >> Hi Dirk, > >> I ran two experiments: > >>=20 > >> **FIRST EXPERIMENT**: > >> Using Matlab software I generated a 600 Hz audio=20 > >> tone and played it through my desktop computer's=20 > >> speakers. > >>=20 > >> I used my TracPhone cell phone to call my home's=20 > >> landline phone number. > >>=20 > >> When my landline phone's answering machine started=20 > >> recording the incoming call I held my TracPhone up to=20 > >> my computer's speakers and recorded the 600 Hz tone=20 > >> on my landline's answering machine. > >>=20 > >> (So, ...that recorded 600 Hz tone has passed through=20 > >> my telephone company's system.) > >>=20 > >> I played the answering machine's audio 'message' and=20 > >> used a microphone to record that message (the 600 Hz=20 > >> tone) on my desktop computer. > >>=20 > >> Then I performed spectrum analysis (64K FFT size) of the=20 > >> microphone/PC recorded audio signal. The spectral result=20 > >> was a "clean" 600 Hz tone with VERY low spectral=20 > >> background noise. > >>=20 > >> **SECOND EXPERIMENT**: > >> Next, using Matlab software I generated a 150 Hz audio=20 > >> tone and played it through my desktop computer's=20 > >> speakers. > >>=20 > >> I repeated the above steps calling my landline phone=20 > >> to record my answering machine's new audio message=20 > >> (now a 150 Hz tone sent through my telephone company's=20 > >> system) on my computer. > >>=20 > >> Again I performed spectrum analysis of the=20 > >> microphone/PC recorded audio signal. The spectral result=20 > >> contained a moderate amount of spectral background noise,=20 > >> but there were two predominant spectral peaks. Those=20 > >> peaks were at large-magnitude spectral peak at 450 Hz=20 > >> and peak at 600 Hz (roughly 12 dB down from the 450 Hz peak). > >>=20 > >> The spectrum contained no measureable energy at 150 Hz. > >>=20 > >> So there you have it: I could transmit a 600 Hz audio=20 > >> tone over my telephone system, but when I transmitted=20 > >> a 150 Hz tone all that arrived at my landline phone=20 > >> was the 2nd and 3rd harmonics (450 and 600 Hz) of the=20 > >> 150 Hz tone. > >>=20 > >> Now here's the neat part: when I listened to my answering=20 > >> machine's "missing 150 Hz" audio it sounded like=20 > >> a 150 Hz tone to me!! > >>=20 > >> SOOooo, the telephone system does NOT work the way=20 > >> I thought it did. It looks like the phone system=20 > >> doesn't transmit my audio voice signal, it only transmits=20 > >> the harmonics of my voice signal. But the destination=20 > >> listener's ear/brain mechanism hears the "missing=20 > >> fundamental" spectral content of my voice. > >>=20 > >> As Al Clark would say, "You really=20 > >> can teach an old dog new tricks." Ha ha. > >>=20 > >> [-Rick-] > > > >Hi Rick, > > > >With the 150 Hz tone, you had distortion in the processing chain generating= > > the harmonics. It could have been a) your signal generation process (migh= > >t be some from poor resampling if that happens, maybe some from your speake= > >rs), b) your transmission process (probably VoIP land line which has audio = > >coding/decoding), c) your recording process (your answering machine probabl= > >y has coding/decoding if it is digital) or d) some combination of a) - c). = > > It would have to happen before the band limiting took place. So you genera= > >ted a third harmonic at 450 Hz and a fourth harmonic at 600 Hz. I suspect t= > >hat there are other harmonics also. If it were not for the distortion, assu= > >ming the same high-pass filter, you would have recorded something very low = > >level or essentially nothing at all. The harmonics from the distortion est= > >ablish the period of the waveform corresponding to the original frequency m= > >aking it sound like you recorded the original tone.=20 > > > >This is different from speech which has the harmonics present when generate= > >d. > > > >Dirk > > I suspect that the system may detect and hard-limit energy below the > cutoff frequency so that harmonics that can be transmitted are forced. > > Otherwise you could get a situation like Rick's experiment where > something (e.g., 150Hz tone) goes in and nothing comes out, which > wouldn't be good. With a hard-limiter on the low frequencies that > problem can be avoided. > > > Eric Jacobsen > Anchor Hill Communications > http://www.anchorhill.com
Eric, I am assuming you mean whether or not audio is present in my response, correct me if I am wrong. I would not think they would want to do that. That could really make a mess of things. For instance if there was power line noise, it could make the noise much more offensive. Dirk
Reply by February 4, 20152015-02-04
On Wednesday, February 4, 2015 at 3:03:11 PM UTC-5, Eric Jacobsen wrote:
> On Wed, 4 Feb 2015 09:28:06 -0800 (PST), bellda2005@gmail.com wrote: > > >On Wednesday, February 4, 2015 at 6:54:45 AM UTC-5, Rick Lyons wrote: > >> On Wed, 28 Jan 2015 12:07:52 -0800 (PST), bellda2005@gmail.com wrote: > >>=20 > >> [Snipped by Lyons] > >> > > >> >Hi Rick, > >> > > >> >You have actually run across the "missing fundamental" 1000's of times b= > >efore. Every time you made a phone call on a land line the fundamental at a= > > minimum was largely suppressed if not effectively eliminated (except maybe= > > for squeaky children and extremely high-pitched people), yet you did not t= > >hink the pitch was any different. The period of a waveform rich with harmon= > >ics will be unchanged whether the fundamental is present or not. That perio= > >d is the pitch, is it not? > >> > > >> >An interesting question (not a proof of anything) is if the fundamental = > >is missing for speech, what do you expect the pitch to be? It hasn't double= > >d because the waveform is not periodic in half the period of the fundamenta= > >l. Some other choice? > >> > > >> >In mobile phones that use spectral coding bandwidth compression it would= > > make sense to eliminate low frequencies including the fundamental to save = > >bits, so I would guess it is probably done there, which could be more 1000s= > > of times you have encountered it. > >> > > >> >Dirk > >> > > >> >Dirk Bell, MSEE, MSCF > >> >Senior-Level DSP Engineer in search of a great opportunity. > >> >Willing to relocate. > >> >https://www.linkedin.com/in/dirkbell > >>=20 > >> Hi Dirk, > >> I ran two experiments: > >>=20 > >> **FIRST EXPERIMENT**: > >> Using Matlab software I generated a 600 Hz audio=20 > >> tone and played it through my desktop computer's=20 > >> speakers. > >>=20 > >> I used my TracPhone cell phone to call my home's=20 > >> landline phone number. > >>=20 > >> When my landline phone's answering machine started=20 > >> recording the incoming call I held my TracPhone up to=20 > >> my computer's speakers and recorded the 600 Hz tone=20 > >> on my landline's answering machine. > >>=20 > >> (So, ...that recorded 600 Hz tone has passed through=20 > >> my telephone company's system.) > >>=20 > >> I played the answering machine's audio 'message' and=20 > >> used a microphone to record that message (the 600 Hz=20 > >> tone) on my desktop computer. > >>=20 > >> Then I performed spectrum analysis (64K FFT size) of the=20 > >> microphone/PC recorded audio signal. The spectral result=20 > >> was a "clean" 600 Hz tone with VERY low spectral=20 > >> background noise. > >>=20 > >> **SECOND EXPERIMENT**: > >> Next, using Matlab software I generated a 150 Hz audio=20 > >> tone and played it through my desktop computer's=20 > >> speakers. > >>=20 > >> I repeated the above steps calling my landline phone=20 > >> to record my answering machine's new audio message=20 > >> (now a 150 Hz tone sent through my telephone company's=20 > >> system) on my computer. > >>=20 > >> Again I performed spectrum analysis of the=20 > >> microphone/PC recorded audio signal. The spectral result=20 > >> contained a moderate amount of spectral background noise,=20 > >> but there were two predominant spectral peaks. Those=20 > >> peaks were at large-magnitude spectral peak at 450 Hz=20 > >> and peak at 600 Hz (roughly 12 dB down from the 450 Hz peak). > >>=20 > >> The spectrum contained no measureable energy at 150 Hz. > >>=20 > >> So there you have it: I could transmit a 600 Hz audio=20 > >> tone over my telephone system, but when I transmitted=20 > >> a 150 Hz tone all that arrived at my landline phone=20 > >> was the 2nd and 3rd harmonics (450 and 600 Hz) of the=20 > >> 150 Hz tone. > >>=20 > >> Now here's the neat part: when I listened to my answering=20 > >> machine's "missing 150 Hz" audio it sounded like=20 > >> a 150 Hz tone to me!! > >>=20 > >> SOOooo, the telephone system does NOT work the way=20 > >> I thought it did. It looks like the phone system=20 > >> doesn't transmit my audio voice signal, it only transmits=20 > >> the harmonics of my voice signal. But the destination=20 > >> listener's ear/brain mechanism hears the "missing=20 > >> fundamental" spectral content of my voice. > >>=20 > >> As Al Clark would say, "You really=20 > >> can teach an old dog new tricks." Ha ha. > >>=20 > >> [-Rick-] > > > >Hi Rick, > > > >With the 150 Hz tone, you had distortion in the processing chain generating= > > the harmonics. It could have been a) your signal generation process (migh= > >t be some from poor resampling if that happens, maybe some from your speake= > >rs), b) your transmission process (probably VoIP land line which has audio = > >coding/decoding), c) your recording process (your answering machine probabl= > >y has coding/decoding if it is digital) or d) some combination of a) - c). = > > It would have to happen before the band limiting took place. So you genera= > >ted a third harmonic at 450 Hz and a fourth harmonic at 600 Hz. I suspect t= > >hat there are other harmonics also. If it were not for the distortion, assu= > >ming the same high-pass filter, you would have recorded something very low = > >level or essentially nothing at all. The harmonics from the distortion est= > >ablish the period of the waveform corresponding to the original frequency m= > >aking it sound like you recorded the original tone.=20 > > > >This is different from speech which has the harmonics present when generate= > >d. > > > >Dirk > > I suspect that the system may detect and hard-limit energy below the > cutoff frequency so that harmonics that can be transmitted are forced. > > Otherwise you could get a situation like Rick's experiment where > something (e.g., 150Hz tone) goes in and nothing comes out, which > wouldn't be good. With a hard-limiter on the low frequencies that > problem can be avoided. > > > Eric Jacobsen > Anchor Hill Communications > http://www.anchorhill.com
Eric, I am assuming you mean whether or not audio is present in my response, correct me if I am wrong. I would not think they would want to do that. That could really make a mess of things. For instance if there was power line noise, it could make the noise much more offensive. Dirk I wouldn't think they would want to do that.
Reply by Eric Jacobsen February 4, 20152015-02-04
On Wed, 4 Feb 2015 09:28:06 -0800 (PST), bellda2005@gmail.com wrote:

>On Wednesday, February 4, 2015 at 6:54:45 AM UTC-5, Rick Lyons wrote: >> On Wed, 28 Jan 2015 12:07:52 -0800 (PST), bellda2005@gmail.com wrote: >>=20 >> [Snipped by Lyons] >> > >> >Hi Rick, >> > >> >You have actually run across the "missing fundamental" 1000's of times b= >efore. Every time you made a phone call on a land line the fundamental at a= > minimum was largely suppressed if not effectively eliminated (except maybe= > for squeaky children and extremely high-pitched people), yet you did not t= >hink the pitch was any different. The period of a waveform rich with harmon= >ics will be unchanged whether the fundamental is present or not. That perio= >d is the pitch, is it not? >> > >> >An interesting question (not a proof of anything) is if the fundamental = >is missing for speech, what do you expect the pitch to be? It hasn't double= >d because the waveform is not periodic in half the period of the fundamenta= >l. Some other choice? >> > >> >In mobile phones that use spectral coding bandwidth compression it would= > make sense to eliminate low frequencies including the fundamental to save = >bits, so I would guess it is probably done there, which could be more 1000s= > of times you have encountered it. >> > >> >Dirk >> > >> >Dirk Bell, MSEE, MSCF >> >Senior-Level DSP Engineer in search of a great opportunity. >> >Willing to relocate. >> >https://www.linkedin.com/in/dirkbell >>=20 >> Hi Dirk, >> I ran two experiments: >>=20 >> **FIRST EXPERIMENT**: >> Using Matlab software I generated a 600 Hz audio=20 >> tone and played it through my desktop computer's=20 >> speakers. >>=20 >> I used my TracPhone cell phone to call my home's=20 >> landline phone number. >>=20 >> When my landline phone's answering machine started=20 >> recording the incoming call I held my TracPhone up to=20 >> my computer's speakers and recorded the 600 Hz tone=20 >> on my landline's answering machine. >>=20 >> (So, ...that recorded 600 Hz tone has passed through=20 >> my telephone company's system.) >>=20 >> I played the answering machine's audio 'message' and=20 >> used a microphone to record that message (the 600 Hz=20 >> tone) on my desktop computer. >>=20 >> Then I performed spectrum analysis (64K FFT size) of the=20 >> microphone/PC recorded audio signal. The spectral result=20 >> was a "clean" 600 Hz tone with VERY low spectral=20 >> background noise. >>=20 >> **SECOND EXPERIMENT**: >> Next, using Matlab software I generated a 150 Hz audio=20 >> tone and played it through my desktop computer's=20 >> speakers. >>=20 >> I repeated the above steps calling my landline phone=20 >> to record my answering machine's new audio message=20 >> (now a 150 Hz tone sent through my telephone company's=20 >> system) on my computer. >>=20 >> Again I performed spectrum analysis of the=20 >> microphone/PC recorded audio signal. The spectral result=20 >> contained a moderate amount of spectral background noise,=20 >> but there were two predominant spectral peaks. Those=20 >> peaks were at large-magnitude spectral peak at 450 Hz=20 >> and peak at 600 Hz (roughly 12 dB down from the 450 Hz peak). >>=20 >> The spectrum contained no measureable energy at 150 Hz. >>=20 >> So there you have it: I could transmit a 600 Hz audio=20 >> tone over my telephone system, but when I transmitted=20 >> a 150 Hz tone all that arrived at my landline phone=20 >> was the 2nd and 3rd harmonics (450 and 600 Hz) of the=20 >> 150 Hz tone. >>=20 >> Now here's the neat part: when I listened to my answering=20 >> machine's "missing 150 Hz" audio it sounded like=20 >> a 150 Hz tone to me!! >>=20 >> SOOooo, the telephone system does NOT work the way=20 >> I thought it did. It looks like the phone system=20 >> doesn't transmit my audio voice signal, it only transmits=20 >> the harmonics of my voice signal. But the destination=20 >> listener's ear/brain mechanism hears the "missing=20 >> fundamental" spectral content of my voice. >>=20 >> As Al Clark would say, "You really=20 >> can teach an old dog new tricks." Ha ha. >>=20 >> [-Rick-] > >Hi Rick, > >With the 150 Hz tone, you had distortion in the processing chain generating= > the harmonics. It could have been a) your signal generation process (migh= >t be some from poor resampling if that happens, maybe some from your speake= >rs), b) your transmission process (probably VoIP land line which has audio = >coding/decoding), c) your recording process (your answering machine probabl= >y has coding/decoding if it is digital) or d) some combination of a) - c). = > It would have to happen before the band limiting took place. So you genera= >ted a third harmonic at 450 Hz and a fourth harmonic at 600 Hz. I suspect t= >hat there are other harmonics also. If it were not for the distortion, assu= >ming the same high-pass filter, you would have recorded something very low = >level or essentially nothing at all. The harmonics from the distortion est= >ablish the period of the waveform corresponding to the original frequency m= >aking it sound like you recorded the original tone.=20 > >This is different from speech which has the harmonics present when generate= >d. > >Dirk
I suspect that the system may detect and hard-limit energy below the cutoff frequency so that harmonics that can be transmitted are forced. Otherwise you could get a situation like Rick's experiment where something (e.g., 150Hz tone) goes in and nothing comes out, which wouldn't be good. With a hard-limiter on the low frequencies that problem can be avoided. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by February 4, 20152015-02-04
On Wednesday, February 4, 2015 at 6:54:45 AM UTC-5, Rick Lyons wrote:
> On Wed, 28 Jan 2015 12:07:52 -0800 (PST), bellda2005@gmail.com wrote: > > [Snipped by Lyons] > > > >Hi Rick, > > > >You have actually run across the "missing fundamental" 1000's of times before. Every time you made a phone call on a land line the fundamental at a minimum was largely suppressed if not effectively eliminated (except maybe for squeaky children and extremely high-pitched people), yet you did not think the pitch was any different. The period of a waveform rich with harmonics will be unchanged whether the fundamental is present or not. That period is the pitch, is it not? > > > >An interesting question (not a proof of anything) is if the fundamental is missing for speech, what do you expect the pitch to be? It hasn't doubled because the waveform is not periodic in half the period of the fundamental. Some other choice? > > > >In mobile phones that use spectral coding bandwidth compression it would make sense to eliminate low frequencies including the fundamental to save bits, so I would guess it is probably done there, which could be more 1000s of times you have encountered it. > > > >Dirk > > > >Dirk Bell, MSEE, MSCF > >Senior-Level DSP Engineer in search of a great opportunity. > >Willing to relocate. > >https://www.linkedin.com/in/dirkbell > > Hi Dirk, > I ran two experiments: > > **FIRST EXPERIMENT**: > Using Matlab software I generated a 600 Hz audio > tone and played it through my desktop computer's > speakers. > > I used my TracPhone cell phone to call my home's > landline phone number. > > When my landline phone's answering machine started > recording the incoming call I held my TracPhone up to > my computer's speakers and recorded the 600 Hz tone > on my landline's answering machine. > > (So, ...that recorded 600 Hz tone has passed through > my telephone company's system.) > > I played the answering machine's audio 'message' and > used a microphone to record that message (the 600 Hz > tone) on my desktop computer. > > Then I performed spectrum analysis (64K FFT size) of the > microphone/PC recorded audio signal. The spectral result > was a "clean" 600 Hz tone with VERY low spectral > background noise. > > **SECOND EXPERIMENT**: > Next, using Matlab software I generated a 150 Hz audio > tone and played it through my desktop computer's > speakers. > > I repeated the above steps calling my landline phone > to record my answering machine's new audio message > (now a 150 Hz tone sent through my telephone company's > system) on my computer. > > Again I performed spectrum analysis of the > microphone/PC recorded audio signal. The spectral result > contained a moderate amount of spectral background noise, > but there were two predominant spectral peaks. Those > peaks were at large-magnitude spectral peak at 450 Hz > and peak at 600 Hz (roughly 12 dB down from the 450 Hz peak). > > The spectrum contained no measureable energy at 150 Hz. > > So there you have it: I could transmit a 600 Hz audio > tone over my telephone system, but when I transmitted > a 150 Hz tone all that arrived at my landline phone > was the 2nd and 3rd harmonics (450 and 600 Hz) of the > 150 Hz tone. > > Now here's the neat part: when I listened to my answering > machine's "missing 150 Hz" audio it sounded like > a 150 Hz tone to me!! > > SOOooo, the telephone system does NOT work the way > I thought it did. It looks like the phone system > doesn't transmit my audio voice signal, it only transmits > the harmonics of my voice signal. But the destination > listener's ear/brain mechanism hears the "missing > fundamental" spectral content of my voice. > > As Al Clark would say, "You really > can teach an old dog new tricks." Ha ha. > > [-Rick-]
Hi Rick, With the 150 Hz tone, you had distortion in the processing chain generating the harmonics. It could have been a) your signal generation process (might be some from poor resampling if that happens, maybe some from your speakers), b) your transmission process (probably VoIP land line which has audio coding/decoding), c) your recording process (your answering machine probably has coding/decoding if it is digital) or d) some combination of a) - c). It would have to happen before the band limiting took place. So you generated a third harmonic at 450 Hz and a fourth harmonic at 600 Hz. I suspect that there are other harmonics also. If it were not for the distortion, assuming the same high-pass filter, you would have recorded something very low level or essentially nothing at all. The harmonics from the distortion establish the period of the waveform corresponding to the original frequency making it sound like you recorded the original tone. This is different from speech which has the harmonics present when generated. Dirk