[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

RE: laptop audio I/O hardware



> From: Jeff Kaiser [mailto:loopersdelight@pfmentum.com] 
>
> I created my live rig in Max/MSP. I don't know how to measure 
> latency, so I can only say I don't notice it after switching to the
MOTU

Technically it is incorrect to say that one audio interface has better
latency than another.  Latency in a computer-based system isn't a fixed
characteristic of a device, it is something you tune for the combination
of components in your system including the audio interface, computer,
operating system, other attached devices, and many other tings.

At a simplistic level, you can think of latency as being defined by
two things: the size of the buffer used by the software device driver
and the size of the buffer used by the hardware digital/audio
converter in the interface.  You have no control over the DAC buffer
size, but you do have control over the driver buffer size.

So assuming that two devices have comparable DAC buffer sizes,
and you set the driver buffers to be the same, the two devices
will have exactly the same amount of latency.  One cannot be better
than another.

"But wait!" I can hear some of you say.  What about those DAC buffers?
If one device has a smaller buffer then it has better latency right?
Technically yes, but we're talking about extremely small buffers,
somewhere between 5 and 100 samples.  Measurements of my sound card
indicate that it has a buffer of around 32 samples.  I don't care how
sensitive a musician you think you are, *you cannot hear this*.

Think you can?  32 samples at 44K is about 0.64 milliseconds.  The
speed of sound is 13.63 inches/millisecond at 75 degrees F.  So a 32
sample buffer produces the same amount of delay as moving your head
about 8.7 inches away from a sound source.  The difference between
slouching over your acoustic guitar vs. sitting upright.

So unless the audio interface's DAC has an extraordinarily large
buffer, latency is determined by the device driver buffer.  As I said
before you get to decide what this is.  You tune latency by setting
the buffer size as low as possible without hearing artifacts that
usually sound like clicks.  How small you can set this buffer is
determined by a lot of things.  One of them is the quality of the
device driver for the audio interface.

If one device has a badly written driver that does not allow you to
select small buffer sizes without hearing clicks, then you could say
that the device "has higher latency" than another but what that
really means is "I cannot tune latency low enough".  

In my limited experience with "prosumer" grade audio interfaces, the
the device driver has never had any effect on latency.  Latency
problems I've had have always been related to the other things running
on the computer, and the other devices attached to it.

So, this was a long winded way of saying that it is difficult to 
make generalizations like "the MOTU has lower latency than the M-Audio".
The latency you can achieve is dependent on many things in the system,
it is rarely an intrinsic characteristic of a single device.

Jeff