| |
Main
Date: 29 Jun 2007 23:48:37
From:
Subject: O.T. How well can you judge times from 1 to 99 milliseconds?
|
Deep in another thread, debate raged briefly (well, sputtered) about
how many milliseconds a chain skip lasts.
I had to nursemaid a client's computer through a tedious install this
afternoon, so I had some time on my hands near another rather ancient
computer and happily scribbled a quick and dirty MS-DOS program that
can be opened or downloaded and run on Windows systems:
http://home.comcast.net/~carlfogel/download/Noises.exe
It's a typically enormous 7k text program that sounds the speaker for
1 to 99 milliseconds, asks you to estimate the time in milliseconds,
and reveals how long it tried to make the noise last.
Three seconds later, it does it again with a new noise.
After 10 guesses--er, after ten increasingly accurate estimates, your
scores--er, results are shown.
You may get better as you play--er, as you train yourself.
This game will--er, this technical tool may give posters some insight
into how well they can estimate short noises, but it says _nothing_
about how long a chain skip lasts.
Like all software from Fogel Labs, the program comes fully guaranteed.
In case of breakage, both halves are yours.
Cheers,
Carl Fogel
|
|
| |
Date: 01 Jul 2007 15:10:48
From: datakoll
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Jul 1, 10:18 am, zencycle <zency...@bikerider.com > wrote:
> On Jun 30, 4:11 pm, carlfo...@comcast.net wrote:
>
>
>
>
>
> > On Sat, 30 Jun 2007 10:51:30 -0700, DirtRoadie <DirtRoa...@aol.com>
> > wrote:
>
> > >On Jun 29, 11:48 pm, carlfo...@comcast.net wrote:
>
> > >> I had to nursemaid a client's computer through a tedious install this
> > >> afternoon, so I had some time on my hands near another rather ancient
> > >> computer and happily scribbled a quick and dirty MS-DOS program that
> > >> can be opened or downloaded and run on Windows systems:
>
> > >>http://home.comcast.net/~carlfogel/download/Noises.exe
>
> > >> It's a typically enormous 7k text program that sounds the speaker for
> > >> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
> > >> and reveals how long it tried to make the noise last.
>
> > >In the interest of digression -
> > >There may be an error introduced due to the coarse resolution of the
> > >clock on an MSDOS system.
> > >http://www.ubr.com/clocks/clock-time-software-for-personal-computers-...
>
> > >I ran into a similar (unexpected) problem last year in creating time
> > >trial timing software for the local club. That wasn't an MSDOS issue,
> > >but it turned out that the timer in the older laptop I was using did
> > >not stay accurate while the software was running. In that case I
> > >created a companion program to read time from a GPS receiver and keep
> > >the computer properly in sync. Later, a newer tablet computer proved
> > >to have a spot-on clock.
>
> > >DR
>
> > Dear DR,
>
> > Beyond that, RonSonic is probably giggling at the thought that cheap
> > computer speakers produce sounds for times precise to the millisecond.
>
> This was my question when I first started reading this thread. PC
> speakers aren't designed for much more than typical voice response
> (~250 - 3000Hz, give or take), and even that isn't very flat (lucky to
> be closer than 3dB in that range) . So, what are the spectral
> components of your 'sound'? Since 1 ms corresponds to 1KHz, all the
> energy would have to be much higher in frequency in order to be useful
> - transient components of a spike would have to be greater than the
> slew of a 1 KHz fundamental. It wouldn't be very 'fair' to use a 100
> Hz tone for the 99 ms stimulus and a 1 KHz tone for the 1 ms stimulus.
> Also, for the sound to be produced faithfully you would need much
> higher frequencies at significanltly higher input amplitudes in order
> to compensate for the reduced response and produce the same sound
> pressure level (dB SPL) that the stimuli in the voice band would
> evoke. From experience, I can tell you that an electrical 1 ms
> envelope will be much longer when acoustically measured with an audio
> spectrum analyzer and audio transducer. Consumer grade speakers just
> aren't that good.- Hide quoted text -
>
> - Show quoted text -
right. analagous to heat blowing off an incandescent filament after
powers off. if your computer has an analog/digital converter you could
record counting digital units. see free software at
www.birds.cornell.edu/raven
|
| |
Date: 01 Jul 2007 07:18:06
From: zencycle
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Jun 30, 4:11 pm, carlfo...@comcast.net wrote:
> On Sat, 30 Jun 2007 10:51:30 -0700, DirtRoadie <DirtRoa...@aol.com>
> wrote:
>
>
>
> >On Jun 29, 11:48 pm, carlfo...@comcast.net wrote:
>
> >> I had to nursemaid a client's computer through a tedious install this
> >> afternoon, so I had some time on my hands near another rather ancient
> >> computer and happily scribbled a quick and dirty MS-DOS program that
> >> can be opened or downloaded and run on Windows systems:
>
> >>http://home.comcast.net/~carlfogel/download/Noises.exe
>
> >> It's a typically enormous 7k text program that sounds the speaker for
> >> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
> >> and reveals how long it tried to make the noise last.
>
> >In the interest of digression -
> >There may be an error introduced due to the coarse resolution of the
> >clock on an MSDOS system.
> >http://www.ubr.com/clocks/clock-time-software-for-personal-computers-...
>
> >I ran into a similar (unexpected) problem last year in creating time
> >trial timing software for the local club. That wasn't an MSDOS issue,
> >but it turned out that the timer in the older laptop I was using did
> >not stay accurate while the software was running. In that case I
> >created a companion program to read time from a GPS receiver and keep
> >the computer properly in sync. Later, a newer tablet computer proved
> >to have a spot-on clock.
>
> >DR
>
> Dear DR,
>
> Beyond that, RonSonic is probably giggling at the thought that cheap
> computer speakers produce sounds for times precise to the millisecond.
>
This was my question when I first started reading this thread. PC
speakers aren't designed for much more than typical voice response
(~250 - 3000Hz, give or take), and even that isn't very flat (lucky to
be closer than 3dB in that range) . So, what are the spectral
components of your 'sound'? Since 1 ms corresponds to 1KHz, all the
energy would have to be much higher in frequency in order to be useful
- transient components of a spike would have to be greater than the
slew of a 1 KHz fundamental. It wouldn't be very 'fair' to use a 100
Hz tone for the 99 ms stimulus and a 1 KHz tone for the 1 ms stimulus.
Also, for the sound to be produced faithfully you would need much
higher frequencies at significanltly higher input amplitudes in order
to compensate for the reduced response and produce the same sound
pressure level (dB SPL) that the stimuli in the voice band would
evoke. From experience, I can tell you that an electrical 1 ms
envelope will be much longer when acoustically measured with an audio
spectrum analyzer and audio transducer. Consumer grade speakers just
aren't that good.
|
| |
Date: 30 Jun 2007 10:51:30
From: DirtRoadie
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Jun 29, 11:48 pm, carlfo...@comcast.net wrote:
> I had to nursemaid a client's computer through a tedious install this
> afternoon, so I had some time on my hands near another rather ancient
> computer and happily scribbled a quick and dirty MS-DOS program that
> can be opened or downloaded and run on Windows systems:
>
> http://home.comcast.net/~carlfogel/download/Noises.exe
>
> It's a typically enormous 7k text program that sounds the speaker for
> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
> and reveals how long it tried to make the noise last.
In the interest of digression -
There may be an error introduced due to the coarse resolution of the
clock on an MSDOS system.
http://www.ubr.com/clocks/clock-time-software-for-personal-computers-pcs/rightime-clock-correcting-program.aspx#DOS
I ran into a similar (unexpected) problem last year in creating time
trial timing software for the local club. That wasn't an MSDOS issue,
but it turned out that the timer in the older laptop I was using did
not stay accurate while the software was running. In that case I
created a companion program to read time from a GPS receiver and keep
the computer properly in sync. Later, a newer tablet computer proved
to have a spot-on clock.
DR
|
| | |
Date: 30 Jun 2007 14:11:04
From:
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Sat, 30 Jun 2007 10:51:30 -0700, DirtRoadie <DirtRoadie@aol.com >
wrote:
>On Jun 29, 11:48 pm, carlfo...@comcast.net wrote:
>
>> I had to nursemaid a client's computer through a tedious install this
>> afternoon, so I had some time on my hands near another rather ancient
>> computer and happily scribbled a quick and dirty MS-DOS program that
>> can be opened or downloaded and run on Windows systems:
>>
>> http://home.comcast.net/~carlfogel/download/Noises.exe
>>
>> It's a typically enormous 7k text program that sounds the speaker for
>> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
>> and reveals how long it tried to make the noise last.
>
>In the interest of digression -
>There may be an error introduced due to the coarse resolution of the
>clock on an MSDOS system.
>http://www.ubr.com/clocks/clock-time-software-for-personal-computers-pcs/rightime-clock-correcting-program.aspx#DOS
>
>I ran into a similar (unexpected) problem last year in creating time
>trial timing software for the local club. That wasn't an MSDOS issue,
>but it turned out that the timer in the older laptop I was using did
>not stay accurate while the software was running. In that case I
>created a companion program to read time from a GPS receiver and keep
>the computer properly in sync. Later, a newer tablet computer proved
>to have a spot-on clock.
>
>DR
Dear DR,
Beyond that, RonSonic is probably giggling at the thought that cheap
computer speakers produce sounds for times precise to the millisecond.
But the computer is probably more accurate than the ear. Most people
will find their guesses falling distinctly on either side of the
predicted times.
(There are strategies for improving your scores over honest guessing,
but I'm sure that no RBT poster would stoop to such underhanded
schemes, at least outside a casino.)
Cheers,
Carl Fogel
|
| |
Date: 30 Jun 2007 10:40:44
From: DirtRoadie
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Jun 29, 11:48 pm, carlfo...@comcast.net wrote:
> Deep in another thread, debate raged briefly (well, sputtered) about
> how many milliseconds a chain skip lasts.
>
> I had to nursemaid a client's computer through a tedious install this
> afternoon, so I had some time on my hands near another rather ancient
> computer and happily scribbled a quick and dirty MS-DOS program that
> can be opened or downloaded and run on Windows systems:
>
> http://home.comcast.net/~carlfogel/download/Noises.exe
>
> It's a typically enormous 7k text program that sounds the speaker for
> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
> and reveals how long it tried to make the noise last.
>
> Three seconds later, it does it again with a new noise.
>
> After 10 guesses--er, after ten increasingly accurate estimates, your
> scores--er, results are shown.
>
> You may get better as you play--er, as you train yourself.
>
> This game will--er, this technical tool may give posters some insight
> into how well they can estimate short noises, but it says _nothing_
> about how long a chain skip lasts.
>
> Like all software from Fogel Labs, the program comes fully guaranteed.
> In case of breakage, both halves are yours.
>
> Cheers,
>
> Carl Fogel
|
| |
Date: 30 Jun 2007 11:22:43
From: Tim McNamara
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
The perception of time is an interesting question, although it's a red
herring in that discussion and shouldn't be dwelt on there, but as a
secondary thread... It's a fascinating topic of which I had read very
little before Carl brought this up (looking in my old textbooks, I find
that it was not addressed at all! It was only briefly mentioned in some
other literature I have). I Googled and scanned through a few articles
as I didn't have time to read them properly today. "Psychoacoustics" is
a good keyword for anyone interested in searching.
Measurements of the perception of time intervals have been done using
difference sensory modes (e.g., using visual, auditory or kinesthetic
stimuli). In auditory perception, humans can discriminate correctly the
sequence between two sounds down to a time interval of 17-20
milliseconds (which is different from what Carl's program apparently
does, which is to ask the subject to estimate the duration of a single
sound. Using a Mac I can't check Carl's program. Maybe a Java
version?). At about 2 milliseconds, the two sounds merge into one.
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1363770
http://ccrma.stanford.edu/workshops/dsp2006/prc/psycho.pdf
Note that the models under discussion in most of these articles are from
a cognitive or information-processing perspective, which assumes that
organisms cannot directly perceive the world but instead perceive
through neural representations of the world. This model has some
strengths and some serious limitations, even though it is still the
current model in most perceptual research. The alternative model,
ecological psychology (James J. Gibson, Eleanor Gibson, et al), assumes
direct perception but is also an incomplete model and has its own
strengths and problems.
In the case of the original thread, we are not talking about perception
of a time interval but a cognitive estimate of a time interval. Jobst
did not say "I heard a sound ten milliseconds long." He estimated the
time interval from reasoning. The sound itself is nearly instantaneous.
I can't remember what you estimated, Carl- that the chain was moving 2.5
mph at the time? If I've done the math correctly, that's 88 half inches
per second. So it takes 0.0114 seconds for the chain to travel one half
inch. Assuming the noise begins when the chain releases from the top of
the tooth until it snaps down into the valley preceding the next tooth
we have an upper limit for the time interval of about 10 ms.
|
| | |
Date: 30 Jun 2007 14:03:55
From:
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Sat, 30 Jun 2007 11:22:43 -0500, Tim McNamara
<timmcn@bitstream.net > wrote:
>Jobst
>did not say "I heard a sound ten milliseconds long." He estimated the
>time interval from reasoning. The sound itself is nearly instantaneous.
>I can't remember what you estimated, Carl- that the chain was moving 2.5
>mph at the time?
[snip]
Dear Tim,
Well . . .
Jobst didn't reason anything. He just claimed that he "is certain" how
long a chain skip takes:
"I am familiar with chain skip and am certain that it occurs in less
than 10ms, or as I said, the sound of a snap of the fingers."
http://tinyurl.com/yr6jps
For all I know, he may be right.
I haven't estimated any times for chain skips.
In fact, that's my question. If we knew how long a half-inch chain
skip takes, that would tell us how fast the short chain section is
going when it bangs into the back of the quick-connect, which in turn
would tell us whether Jobst's elegant rubber-band test is roughly
right or grossly over-powered.
I did calculate how long a half-inch chain skip takes at free-fall
acceleration: 0.051 seconds.
But as I pointed out, free-fall is probably a bad example. The frame,
leg muscles, and (probably) the chain release stored energy, like
springs, and therefore probably accelerate the chain faster than
gravity.
Here's a table that I just posted back in that thread:
avg final final ratio
speed speed accel speed of
ms fps fps f/s^2 mph impact
0.001 41.667 83.333 83,333.33 56.8 3226.8
0.002 20.833 41.667 20,833.33 28.4 806.6
0.005 8.333 16.667 3,333.33 11.5 132.3
0.010 4.167 8.333 833.33 5.7 32.5
0.020 2.083 4.167 208.33 2.8 7.8
0.040 1.042 2.083 52.08 1.4 2.0
0.051 0.817 1.634 32.04* 0.8 0.6
*free fall = 32 f/s^2
With a rubber band, you can fling a couple of chain links about 50
feet in about a second, as I did, which places a rubber band test near
the top of the table.
But the range of impact from 0.001 to 0.010 seconds ia hundred-fold,
from an impact of 32 X to an impact of 3,200 X.
"Nearly instanteous" just isn't good enough to describe the outcome.
Cheers,
Carl Fogel
|
| |
Date: 30 Jun 2007 11:17:22
From: datakoll
Subject: Re: O.T. How well can you judge times from 1 to 99 milliseconds?
|
On Jun 30, 1:48 am, carlfo...@comcast.net wrote:
> Deep in another thread, debate raged briefly (well, sputtered) about
> how many milliseconds a chain skip lasts.
>
> I had to nursemaid a client's computer through a tedious install this
> afternoon, so I had some time on my hands near another rather ancient
> computer and happily scribbled a quick and dirty MS-DOS program that
> can be opened or downloaded and run on Windows systems:
>
> http://home.comcast.net/~carlfogel/download/Noises.exe
>
> It's a typically enormous 7k text program that sounds the speaker for
> 1 to 99 milliseconds, asks you to estimate the time in milliseconds,
> and reveals how long it tried to make the noise last.
>
> Three seconds later, it does it again with a new noise.
>
> After 10 guesses--er, after ten increasingly accurate estimates, your
> scores--er, results are shown.
>
> You may get better as you play--er, as you train yourself.
>
> This game will--er, this technical tool may give posters some insight
> into how well they can estimate short noises, but it says _nothing_
> about how long a chain skip lasts.
>
> Like all software from Fogel Labs, the program comes fully guaranteed.
> In case of breakage, both halves are yours.
>
> Cheers,
>
> Carl Fogel
frequency?
is the transducer ordered to stop?
|
|
