Benchmark DAC1 now available with USB

[EliasGwinn]

Yes, kmixer itself is relatively harmless. As for iTunes, you must be careful of a few things:

1. iTunes volume control should always be set to 100%, as volume reduction in iTunes causes severe distortion. This is because volume calculations will result in 24-bit words, even if the audio was initially 16-bit (due to remainders after division). iTunes will then truncate to 16-bits, instead of dithering or simply passing 24-bits

2. iTunes will convert the sample-rate of any audio with different sample-rates then that set in QuickTimes preferences (yes, QuickTime preferences affects iTunes!!) Be sure that sample-rate corresponds to the sample rate of the audio you are listening to.

3. For the purest audio playback, do not use the "sound check" or "sound enhancer" features in iTunes. All of these DSP and/or audio plug-ins in all media players should be avoided to obtain faithful playback.

All of this will be covered in our "Guide to Computer Audio", which will be available on our website soon. I'll let you know...

[EliasGwinn]

As for ASIO and/or kernel streaming, we haven't tested to prove or disprove specific performance of these technologies. We tested USB audio devices which used the ASIO protocol, and we were not satisfied with their performance. This could likely be the programmers, not the protocol (API). Nonetheless, as we discovered the potential that was inherent with native usbaudio.sys and kmixer, plus the added convenience of not having to deal with 3rd party drivers which may "argue" with other drivers on your computer, etc, we decided that "native" was the way to go.

[EliasGwinn]

Originally Posted by Andrew_WOT

Isn't that the case for Bulk transfer type only, and not Isochronous?

This is theoretically true, but USB wasn't really built for steady streaming, even in Isochronous. When you watch the stream on a bus analyzer, it becomes apparent.

I think what happens is...USB activity is sort of a queued process, and follows the queue with relative prorioty, etc. Therefore, when other activity takes priority, USB activity is compromised.

I wish I knew with certainty, but even the most official publications on this will disagree with each other!!

Nonetheless, USB audio does suffer from 'ticks' and drop outs, which, for whatever reason, is due to insufficient streaming capabilities. When developing the DAC1 USB, we put several "checks" and buffers into place to monitor the stream and prevent these errors from occuring. We played audio through the DAC1 USB while taxing the processor of the computer with other apps, etc, and we couldn't get the DAC1 USB to tick or pop.

[EliasGwinn]

Originally Posted by greenleaf

awsome Foobar will be included, for sure?

Yes, Foobar will be included.

In fact, we were very impressed with Foobar. It has a lot of really neat features that are specifically designed for faithful playback.

[5Kurt]

Originally Posted by EliasGwinn

This is theoretically true, but USB wasn't really built for steady streaming, even in Isochronous. When you watch the stream on a bus analyzer, it becomes apparent.

Hi Elias, thanks for the very valuable information.

What you have said for USB, can we assume the same for Firewire connections?

[EliasGwinn]

Originally Posted by 5Kurt

Hi Elias, thanks for the very valuable information.

What you have said for USB, can we assume the same for Firewire connections?

No, Firewire protocol is completely different. I don't know too much about Firewire yet, so I can't really tell you how they are different.

[gregeas]

Elias:

Thanks for all of the valuable information here.

Do you have any observation on the quality of USB audio versus the quality of a good CD transport? I assume that as long as the bits are the identical, it all comes down to jitter. Can USB deliver a better signal to the DAC1 than the best transports?

[5Kurt]

Elias :

For people who already own a DAC-1 (not the USB version) and use PC as transport do you recommend SPDIF or Optical input?

Thanks

ps: sorry we ask you lot of things but we rarely find an engineer to ask such questions.

[The Monkey]

Also, Elias, what are your primary 'phones?

[wang228]

Originally Posted by EliasGwinn

Here is a summary of what kmixer does to audio:

CONDITION 1: One audio stream is sent through audio through kmixer at a time

RESULT: kmixer streams the audio bit-transparently - that is, bit-for-bit, what goes in, also comes out. We have tested and proven this using a test function called 'Bittest' by Audio Precision.

CONDITION 2: Two audio streams of same sample rate are sent through kmixer

RESULT: kmixer streams both without problems, ASSUMING THE SUM OF THE AUDIO STREAMS DOES NOT ECLIPSE 0 dBFS!! Just like any digital mixer, if the sum of the audio eclipses 0 dBFS, digital clipping will occur, which is not popular among audio enthusiasts. However, if it does not eclipse 0 dBFS, there should be no problems. This was confirmed by playing a 'Bittest' stream with one app and a silence (all 0's) stream with another. The result was bit-transparency. NOTE: When multiple audio streams are summed in kmixer (even 16-bit audio streams), the result will be a 24-bit audio stream. THIS IS WHAT WE WANT, assuming we have a 24-bit device to recieve it.

Aah.. thank you. Finally things start to make sense for me why my dac indicates 44.1K sampling rate from USB even without bypassing kmixer. Somebody suggested it was the computer's unreliable clock, but i have been skeptical.