CD playback speed accuracy

[[email protected]]

I just did some tests to evaluate the speed accuracy of the playback drives
around here.

Playing a wav file on a computer uses the computer's clock but playing an
audio disk on a computer drive or standalone/offline player uses the
drive's own clock.

Plextor 48/24/48a... 0.0004 % error
Plextor 40Max... 0.2 % error (shameful)
Denon standalone... 0.01% error
Panasonic Walkman... 0.24% error
....

(The low error on the first Plextor makes me suspicious that it somehow
used the computer's clock.)

I had hoped for better. Oh well.

[Arny Krueger]

"Sugarite" > wrote in message
...

> wrote in message
...

> > Playing a wav file on a computer uses the computer's clock but playing
an
> > audio disk on a computer drive or standalone/offline player uses the
> > drive's own clock.

Maybe yes, maybe no. (see below)

> Playing a CD in a computer's CD drive does not use the drive's clock, not
> since Windows and MacOS started using the drive interface (IDE, SCSI)
> instead of the analog or s/pdif outputs. I'm not sure when Windows
started
> doing that,

Optional in Win98SE, but I believe its been the default since. Prior to
Win98SE the OS did things the old fashioned way, but there were many music
players that used DAE to play audio discs. Recent versions of the Windows
Media Player also default to using SAE to play audio discs.

> but on the Mac it was over 3 years ago. The audio data is
> essentially extracted to a buffer and played from there using the
> soundcard's clock. I haven't hooked up the audio cables for a CD drive
for
> my past 4 computers.

Using DAE to play audio CDs is also handy for machines with more than one
device that is capable of playing them. Most sound cards had only one input
for CD analog audio, and most audio production cards have none.

> How are you measuring this anyway?

One way to avoid the procedure you correctly describe would be to use a
Win98SE system and ensure that the relevant option is turned off, which BTW
default to off. Then, you have to avoid using a recent version of the
Windows Media Player or any number of other audio CD-playing software that
use the procedure you describe anyway.

The usual tolerance for subjectively-perfect pitch is something like 0.5%,
so the 4 players that were tested are all good enough for playing music to
listen to. The only one that was close wasn't a computer drive.

Of course if you want to do production work and keep things in synch for
long periods of time, the ideal solution is to use the one clock for
everything.

[[email protected]]

"Arny Krueger" > wrote:

>The usual tolerance for subjectively-perfect pitch is something like 0.5%

I agree in the broad sense. But, oddly, when the playback is very close to
true we can hear *extremely* close offsets. To avoid pointless
disagreements, I won't cite numbers.

[[email protected]]

"Sugarite" > wrote:

>How are you measuring this anyway?

I burned an audio disk with just 2 one-bit clicks about 4 minutes apart
with a few seconds of silence outside the interval to avoid frame-rounding.

The playback output was fed from the front-panel analog jack to the
soundcard and captured by CoolEdit to a file. The resulting wav file was
trimmed to the clicks and its sample count was divided by the original
count.

I'm running win98se but this shouldn't matter.

[DaveDrummer]

Maybe cooledit was off (though I doubt it)

> wrote in message
...
> "Sugarite" > wrote:
>
> >How are you measuring this anyway?
>
> I burned an audio disk with just 2 one-bit clicks about 4 minutes apart
> with a few seconds of silence outside the interval to avoid
frame-rounding.
>
> The playback output was fed from the front-panel analog jack to the
> soundcard and captured by CoolEdit to a file. The resulting wav file was
> trimmed to the clicks and its sample count was divided by the original
> count.
>
> I'm running win98se but this shouldn't matter.

[Arny Krueger]

> wrote in message

> "Arny Krueger" > wrote:
>
>> The usual tolerance for subjectively-perfect pitch is something like
>> 0.5%
>
> I agree in the broad sense. But, oddly, when the playback is very
> close to true we can hear *extremely* close offsets. To avoid
> pointless disagreements, I won't cite numbers.

I agree that the issue of JND of pitch is far more complex issue than a
simple number like 0.5%.

[Andy Heninger]

"Arny Krueger" > wrote
> > Playing a CD in a computer's CD drive does not use the drive's clock,
not
> > since Windows and MacOS started using the drive interface (IDE, SCSI)
> > instead of the analog or s/pdif outputs.

> Optional in Win98SE, but I believe its been the default since.

The option for CD playback using the analog connnection between the
drive and the computer still exists in Windows. In Windows Media Player,
select tools menu->Options. In the resulting dialog box, select the
"devices" tab. Select the CD drive, click the properties button.
In the properties dialog, select the "analog playback" button.
That's it. If it doesn't work, check that CD input on the volume
control isn't muted or all the way down, and that the audio cable
between the drive and the sound card or mother board is there.

This is on XP with Media Player 9.

-- Andy Heninger