Images of the piano waveform are here
http://www.pbase.com/filmworks4ever/0p44percent_and_0p88_percent
Well, I wanted to find out how truly poor my sense of hearing was. I had
been keenly following some posts on [ARSCLIST] and I was positive that
my sense of hearing was not even remotely close to detect things that
should have been obvious. So I decided to test myself.
O Would a pitch shift of 0.44% make a singer sound like a canary?
O Would a pitch shift of 0.44% be obvious?
O Is it easier to detect a pitch shift with a real instrument as opposed
to a sine wave test tone?
Step 1 – I took 3 seconds of piano, pitch shifted it 0.88% just to be
generous and glued it on the back. Then I played the material in an
endless loop.
http://www.bitworksfilm.com/AudioSamples/0p88PercentPianoLoop.wav
Step 2 - I repeated step 1 using a 400 hz tone and a 0.5% shift
http://www.bitworksfilm.com/AudioSamples/0p5Percent400hzLoop.wav
Step 3 - I repeated step 1 using a 1000 hz tone and a 0.5% shift
http://www.bitworksfilm.com/AudioSamples/0p5Percent1000hzLoop.wav
Observations and testing hints
· I used Audacity Transport-Loop after selecting the extent of the test
waveform so that it would loop smoothly.
· Use enclosed headphones.
· If you think you can hear the change in pitch, close your eyes for 30
seconds while listening so you can not tell when you cross the boundary
from one tone to the next. And see if you are still sure you can hear
the difference.
How bad is my hearing?
· On the 3 seconds of piano shifted by 0.44% times two (0.88%) I could
not reliably hear the difference. I found the fact there was a richer
timbre made it much more difficult to hear a pitch shift
· I experienced absolutely nothing that would make any vocalist sing
like a canary.
· With the 1000 Hz 1005 hz tones I could hear what I would describe as a
slide or bump as the second tone started, but once into the second tone
I was not really aware of a new frequency.
· With the 400 Hz 402 hz tones hearing the shift from one to the other
was much less obvious.
· I thought I could hear the frequency shift in the test tones if I
listened really intently, but then when I closed my eyes so that I was
unaware of where the playback was, I was less certain. Hearing the
transition between the 1000 1005 was somewhat apparent.
· In fact, after sitting calmly for a long time I could experience a
tiny wavering in frequency that did not exist in the source, but
probably had more to do with my pulse. Enclosed headsets required and
keep it looping.
· If I clenched and unclenched my jaw while listening to the 0.5%
shifted test tones I could cause a subjective pitch shift much greater
than I could detect while sitting relaxed.
· I also felt my pulse on my wrist and could, if I was in a suggestive
mode, convince myself that my heart-beat could be detected as a
pitch-shift in my hearing. I should do some heavy exercises and retest
this. And we won’t even bring yawning into this. Also, try standing on
an old mechanical un-dampened weigh scale and watch the needle move with
your pulse (nothing to do with this test or audio – or does it?)
Conclusions on my own sense of hearing
1) An absolute pitch shift of 0.44% is all but undetectable to me in any
reliable way, especially at 400 hz and this is when the tones are played
back to back.
Re-entering a room would make any possible difference absolutely
un-detectable to me.
2) Conducting the test with material with richer timbre made it much
harder, not easier to detect a pitch shift.
3) None of the things that were described as obvious were obvious to me
at all.
If your hearing is as poor as mine then a recording could be off many
times 0.44% and you would not have a blessed clue.
I could never depend on my own hearing to know the playback speed to a
0.44% tolerance.
The suggested piano test (see below) seems like a great way to check a
recording.
Regarding the quote “Any musician must be able to discern such a
difference. Otherwise, they'd never be able to tune their instruments
(or voices) to their fellow performers. An orchestra with half its
members that far out of tune would be painful to hear.”
How is this known?
Has this ever been tested after an orchestra has tuned?
0.44% is a very small amount.
Painful to hear?
Has anyone measure whether the tuning of a SAX, trumpet, or clarinet
varies with sound intensity or stays constant even? I strongly suspect
piano string change their tuning with the intensity of the note, as real
life is not linear.
Furthermore, tuning an orchestra to a reference note, or hitting piano
keys to detect the correct playback frequency of a recording (again a
great suggestion) has nothing to do with being able to detect an
absolute pitch error. What we humans detect when we do this is a
difference between the reference subject sound and the reference tone.
This difference can vary an order of magnitude while you tune while
subject tone itself changes by less that a percent.
Quotes from the posts [ARSCLIST] than intrigued me
Start of Quote –
Any musician must be able to discern such a difference. Otherwise,
they'd never be able to tune their instruments (or voices) to their
fellow performers. An orchestra with half its members that far out of
tune would be painful to hear.
And, any experienced listener, especially those who have spent a lot of
time finding the correct playback speeds for historical recordings, will
be able to tell the difference. Easily. It's not just a matter of pitch,
but of timbre, as well. A 0.44% error in speed causes a significant
change in the timbre of a singer's voice. Pitch Geraldine Farrar 0.44%
high and she turns into a canary. Ditto Nelly Melba.
And, the difference is much easier to discern with real musical program
material than it would be with test tones.
End of Quote
Start of Quote –
My comment on the change in timbre does, indeed, relate to a record
played at a 0.44% shifted speed. And, yes, a change of that magnitude
will distort a singer's timbre and that change will be easily heard by a
trained ear. I pretty sure Jyyy Hyyyy will agree with this, but I'll let
him weigh in for himself.
Like I said, the difference on test tones may be difficult to detect,
but on program material it will be audible. On test tones, there's no
change in timbre, and a change in timbre is an important part of what
this is about.
-End of Quote
Start of Quote –
It depends on what it is that is playing. I don't have absolute perfect
pitch, so for a test tone, no, I wouldn't know that. But with most
recordings of singers,especially singers you are somewhat familiar with,
the difference would be readily apparent upon walking into a room.
Orchestral string tone is another fairly good telltale sign. Thin
screechy strings are often sharp, and fat dull sounding ones can
indicate flat. It is harder to hear initially with something like a
piano, where tones don't have much vibrato and behave about the same at
correct pitch and close-but-erroneous pitch.
-End of Quote
Start of Quote –
And yes, I agree with Gxxx's posts. After pitching things for many
years, you develop some kind of "sixth sense" where you hear something
and say to yourself, oops, I better check the pitch of that. This is not
a reliable test, of course, just a hunch that hits you involuntarily.
Most of the time, the hunch is correct. But of course I check everything
anyway, to be sure.
Because I am "piano keyboard friendly," I have found that playing along
with something, just by ear, on my electronic keyboard (that unlike me,
does have true perfect pitch), the adjustment that needs to be made
becomes very obvious, even where it is really subtle. When I do this, I
am startled by how often my hand just lands on the right pitch for
something, unconsciously, even though I do not have perfect pitch. I
often wonder if I would have developed perfect pitch if I had seriously
studied the violin as a child, as so many string players have it. It is
definitely something that can be learned, probably starting at the right
age, as too many string players have it. However, most people with
perfect pitch can be fooled by near but not exact pitch. So even they
would still have to check things, just as I do.
I would encourage anyone who is responsible for pitching recordings to
try to electronic keyboard approach. You really don't need any kind of
well developed piano ability to do this. I think almost anyone could do
it well enough to pitch things.
-End of Quote
On 5/1/2017 1:42 PM, Gary A. Galo wrote:
> My comment on the change in timbre does, indeed, relate to a record played at a 0.44% shifted speed. And, yes, a change of that magnitude will distort a singer's timbre and that change will be easily heard by a trained ear. I pretty sure John Haley will agree with this, but I'll let him weigh in for himself.
>
> Like I said, the difference on test tones may be difficult to detect, but on program material it will be audible. On test tones, there's no change in timbre, and a change in timbre is an important part of what this is about.
>
> Gary
>
> ____________________________
>
>
> -----Original Message-----
> From: Association for Recorded Sound Discussion List [mailto:[log in to unmask]] On Behalf Of John Gledhill
> Sent: Monday, May 01, 2017 1:29 PM
> To: [log in to unmask]
> Subject: Re: [ARSCLIST] How many half-tones from 78 rpm to 80 rpm - really
>
> Fascinating, however your example does not relate to a record playing at a 0.44% shifted speed.
>
> When you tune an instrument, your can compare a reference note to your note and listen for the beat frequency. This is entirely a different matter. I can certainly hear the beat frequency between 1000 hz and 1004 hz being 4 hz.
>
> My point is that if I stop either the 1000 hz or the 1004 hz tone, I find it hard to tell which has stopped. And this is without leaving the room and coming back or dealing with a non test tone (live material).
>
> Regarding "Pitch Geraldine Farrar 0.44% high and she turns into a canary. Ditto Nelly Melba." Are you saying that shifting less than 1 /13th of a semi-tome is not only detectable but renders a normal voice like a canary?
>
>
>
>
> On 5/1/2017 12:58 PM, Gary A. Galo wrote:
>
>> Any musician must be able to discern such a difference. Otherwise, they'd never be able to tune their instruments (or voices) to their fellow performers. An orchestra with half its members that far out of tune would be painful to hear.
>>
>> And, any experienced listener, especially those who have spent a lot of time finding the correct playback speeds for historical recordings, will be able to tell the difference. Easily. It's not just a matter of pitch, but of timbre, as well. A 0.44% error in speed causes a significant change in the timbre of a singer's voice. Pitch Geraldine Farrar 0.44% high and she turns into a canary. Ditto Nelly Melba.
>>
>> And, the difference is much easier to discern with real musical program material than it would be with test tones.
>>
>> Gary
>>
>>
>> -----Original Message-----
>> From: Association for Recorded Sound Discussion List
>> [mailto:[log in to unmask]] On Behalf Of John Gledhill
>> Sent: Monday, May 01, 2017 12:28 PM
>> To: [log in to unmask]
>> Subject: Re: [ARSCLIST] How many half-tones from 78 rpm to 80 rpm -
>> really
>>
>> Are you saying that you could walk in and out of a room with material playing and tell the differences between something that had been shifted by 0.44%. Such as music or a 1000 hz versus a 1004 hz tone?
>> And, describe the difference as dramatic?
>> I find this absolutely fascinating, being somewhat tone deaf myself.
>>
>> I assume we are not talking about wow or flutter but absolute pitch.
>>
>> On 5/1/2017 10:56 AM, John Haley wrote:
>>
>>
>>> Thanks, Michael. The .44 pitch speed/pitch error is a quite
>>> significant one, even a dramatic one, in terms of its effect on music.
>>>
>>> This brings me back to the position I stated earlier in this string,
>>> that while all the science is dandy, you should be checking pitch
>>> individually for every 78 record, at least for dubbing work. Without
>>> research you sometimes can't tell where a record was recorded (which
>>> can be different from the place of manufacture, as Michael pointed
>>> out), whereas the pitch is manifest from the record itself and easily
>>> ascertained. Using all the scientific aids is helpful but doing that
>>> without actually checking the pitch is ultimately going to lead to errors.
>>>
>>> And just checking the pitch is way faster than researching where a
>>> record was recorded and then doing all the involved math.
>>>
>>> While this is off-topic, checking the pitch when dubbing LPs is also
>>> a great idea. LP pitch is much more standard than 78 pitch, but
>>> still not totally reliable. So many different companies, employing
>>> so many human beings, created records, over a long period of time.
>>> Nothing is truly "standard."
>>>
>>> E.g., in past decades, Decca/London Records deliberately released
>>> some opera recordings way off pitch, just to save record space and
>>> jam an opera onto two records instead of three. Of course that
>>> should never have happened at a reputable, big company, but human
>>> beings running the company made recognizable human decisions.
>>> Completely erroneous and misguided, but there we are.
>>>
>>> Best,
>>> John
>>>
>>>
>>>
>>>
>>> On Mon, May 1, 2017 at 10:13 AM, Michael Shoshani<
>>> [log in to unmask]> wrote:
>>>
>>>
>>>
>>>
>>>> Hi John
>>>>
>>>> Not so much manufactured as recorded. Strobe discs are dependent on
>>>> a specific number of lines, which vary according to the frequency of
>>>> the blinking of the light source. American 60Hz strobe discs require
>>>> 92 bars, which gives a speed of 78.26 RPM, but in countries where
>>>> 50Hz is the power frequency, the strobe discs require 77 bars, which
>>>> gives a speed of 77.92 RPM. These are thus the speeds at which
>>>> electrically driven record players were factory calibrated, and thus
>>>> the speeds at which the record companies in their respective areas recorded their material.
>>>>
>>>> Record companies on both sides of the Atlantic regularly exchanged
>>>> material; Jack Hylton's HMV records were issued on Victor here, for
>>>> example, and Duke Ellington's Victor records were issued on HMV
>>>> there. But these were master pressings, not dubbed and
>>>> speed-adjusted. Presumably the 0.44% difference in speed is either
>>>> unnoticed or tolerated by most listeners, since the playing
>>>> equipment for each would be slightly incorrect for the other country.
>>>>
>>>> Michael Shoshani
>>>> Chicago
>>>>
>>>>
>>>> On Mon, May 1, 2017 at 8:58 AM, John Haley<[log in to unmask]> wrote:
>>>>
>>>>
>>>>
>>>>
>>>>> Michael, are you saying that 78s manufactured in Europe play at a
>>>>>
>>>>>
>>>>>
>>>> different
>>>>
>>>>
>>>>
>>>>> correct speed than 78's manufactured in the US? I have never heard
>>>>> anything like that before.
>>>>>
>>>>> Best,
>>>>> John
>>>>>
>>>>>
>>>>>
>>>>> On Mon, May 1, 2017 at 9:48 AM, Michael Shoshani<
>>>>> [log in to unmask]
>>>>>
>>>>>
>>>>>
>>>>>> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>>> Hi Gary,
>>>>>>
>>>>>> While the KAB Speed Strobe works independently of the power line
>>>>>>
>>>>>>
>>>>>>
>>>>> frequency,
>>>>>
>>>>>
>>>>>
>>>>>> and thus works equally well in 60 Hz and 50 Hz countries in
>>>>>> theory, I
>>>>>>
>>>>>>
>>>>>>
>>>>> would
>>>>>
>>>>>
>>>>>
>>>>>> submit that in practice it is still geared to the speeds provided
>>>>>> on 60
>>>>>>
>>>>>>
>>>>>>
>>>>> Hz
>>>>>
>>>>>
>>>>>
>>>>>> turntables, which means that electrically recorded 78s from Europe
>>>>>> will
>>>>>>
>>>>>>
>>>>>>
>>>>> be
>>>>>
>>>>>
>>>>>
>>>>>> off.
>>>>>>
>>>>>> Its specs indicate 78.26 for 78rpm, which is the 60Hz standard;
>>>>>> Electrically recorded 78s from countries where 50Hz is the power
>>>>>>
>>>>>>
>>>>>>
>>>>> frequency
>>>>>
>>>>>
>>>>>
>>>>>> are recorded at 77.92 RPM. Anyone in the UK or Europe, for
>>>>>> example,
>>>>>>
>>>>>>
>>>>>>
>>>> who
>>>>
>>>>
>>>>
>>>>>> uses the SpeedStrobe to set their turntables at 78, will be
>>>>>> playing
>>>>>>
>>>>>>
>>>>>>
>>>> their
>>>>
>>>>
>>>>
>>>>>> locally manufactured records at a speed 0.44% faster than they
>>>>>> should
>>>>>>
>>>>>>
>>>>>>
>>>> be
>>>>
>>>>
>>>>
>>>>> -
>>>>>
>>>>>
>>>>>
>>>>>> a slightly greater pitching error than the 0.42% Caruso one. (The
>>>>>> SpeedStrobe does not offer 77.92, which seems an amazing oversight
>>>>>> for
>>>>>>
>>>>>>
>>>>>>
>>>> a
>>>>
>>>>
>>>>
>>>>>> product intended for worldwide use.)
>>>>>>
>>>>>> Michael Shoshani
>>>>>> Chicago
>>>>>>
>>>>>>
>>>>>> On Thu, Apr 27, 2017 at 2:40 PM, Gary A. Galo<[log in to unmask]>
>>>>>>
>>>>>>
>>>>>>
>>>>> wrote:
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>
>>>>>>> Hi George,
>>>>>>>
>>>>>>> Following Aida Favia-Artsay, the difference between 76.60 (60Hz)
>>>>>>> and
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> 76.92
>>>>>>
>>>>>>
>>>>>>
>>>>>>> (50Hz) is 0.42%. This is an unacceptable pitching error. Are you
>>>>>>>
>>>>>>>
>>>>>>>
>>>>> telling
>>>>>
>>>>>
>>>>>
>>>>>> me
>>>>>>
>>>>>>
>>>>>>
>>>>>>> that every 76.6-rpm Caruso record pitched using her 50 Hz strobe
>>>>>>> will
>>>>>>>
>>>>>>>
>>>>>>>
>>>>> be
>>>>>
>>>>>
>>>>>
>>>>>>> 0.42% off?
>>>>>>>
>>>>>>> It would seem that, for turntables lacking a digital readout, a
>>>>>>>
>>>>>>>
>>>>>>>
>>>>> sensible
>>>>>
>>>>>
>>>>>
>>>>>>> solution is KAB's Speed Strobe:
>>>>>>>
>>>>>>> http://www.kabusa.com/strobe.htm
>>>>>>>
>>>>>>> The Speed Strobe comes with its own LED lamp, which is
>>>>>>> illuminated
>>>>>>>
>>>>>>>
>>>>>>>
>>>>> with a
>>>>>
>>>>>
>>>>>
>>>>>>> quartz-locked AC signal. Therefore, it is not dependent on the
>>>>>>> power
>>>>>>>
>>>>>>>
>>>>>>>
>>>>> line
>>>>>
>>>>>
>>>>>
>>>>>>> frequency, and will work equally well in 60 Hz and 50 Hz countries.
>>>>>>>
>>>>>>> Gary
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>> --
>> John Gledhill
>> BIT WORKS Inc.
>> 905 881 2733
>> [log in to unmask]
>> www.bitworks.org
>>
>>
> --
> John Gledhill
> BIT WORKS Inc.
> 905 881 2733
> [log in to unmask]
> www.bitworks.org
>
--
John Gledhill
BIT WORKS Inc.
905 881 2733
[log in to unmask]
www.bitworks.org
|