XMOS-based Asynchronous USB to I2S interface

[soundcheck]

Originally Posted by Wolfsin;3016263Please check out audiodesign's thread "USB to I2S: M2Tech vs XMOS vs Open-source Qnktc" for listening tests.

The beauty of the "Lucian Solution" is its broad applicability. Getting the full benefit, i.e. bypassing the isolator, requires external power. Even including the added jitter of the isolator it sounds VERY good. The simple message of how wonderful it is has been garbled in this thread by perfectionist musings IMO.

It is hard to 'guarantee' anything as complicated as a homebrew setup. I chose to test with a nearly ten year old version of Windows on a variety of hardware, some quite old. The DPClat program tells you everything you need to know for that platform. For Linux you might need to download the Audio Class 2 package and install it for older distros.

I do not doubt that the WavIo perfoms better then others. Though it's all relative.
All discussions, feedback and reviews apply for a single person with a unique system only - always.

By looking at the board it really seems that Lucien did quite a good job. I've been in contact with Lucien via mail earlier.

However. I did not pull the trigger.

Meanwhile (after years of being around in computer based audio) all kind of "big promise" interfaces and DACs are piling up in my workshop.

To me all this stuff becomes commodity, once you focus on certain key characterisics. Differences are becoming neglectable.


The key differentiator and key buying criteria for my next interface is how much that interface is able to cope with the input data and distortions without the need for a custom driver.

I don't want to tweak any transport or any OS or network to get the best sound from my audio interface anymore. After 5 years of actively being in the tweaking business, I can say it'll never end, unless the audio interfaces get the job done properly.

Todays computer based audio is still like gambling. Every HW change or every SW change/update/upgrade can change your audio experience significantly.

The only way to get around this mess is to get the audio interfaces done right.

If I read comments like Amarra sound better then Pure Music. Linux sounds better then Windows 7 (assuming bit transparent data transfer). Using Fidelizer lifts the sound up. Wasapi exclusive sounds best. Asf. Asf.

I'll step on the break.

Even manufacturers give you weired advise which actually show the weaknesses of their product: on e.g MHDT USBridge product page you'll find following advise "Since Windows® is not a real time system, so the lower DPC latency value will result the better sound quality. "

I'd never buy that interface.

To me the goal really is to hook up any transport to the audio interface and end up with the same sound experience.


Unfortunately I havn't read any trustworthy review or thread or product comparision, which would come up with some kind of TIR (Transport Immunity Ratio) evalution.

[Ryelands]

Originally Posted by soundcheck

The only way to get around this mess is to get the audio interfaces done right . . . To me the goal really is to hook up any transport to the audio interface and end up with the same sound experience.

Well, who's going to dispute that? But the facts are that we haven't got near that goal, that even those who develop and implement the likes of asynch protocols for USB audio admit that cables, OS configuration and similar variables can and do affect the sound and that where to go from here is by no means clear due to what John Swenson usefully called "second order effects" - see e.g.

Comparisons with other interfaces - John Swenson - Computer Audio Asylum

How can USB performance impact audio quality? - John Swenson - Computer Audio Asylum

Meanwhile, those of us who want to listen to our music libraries today and tomorrow, not only at some remote point in the future when designers have fixed issues they cannot currently identify let alone resolve make do with what we've got.

You chose to abandon an earlier "must have" route - a low-powered Linux box - and chose instead to modify the Squeezebox with, according to reports on the forums, considerable success. So I'm slightly surprised to see you looking anew at a PC interface before deciding publicly that it has little to offer.

Incidentally, I may well have missed the argument but I'm curious why you lay such stress on an interface not needing a "custom" driver. I'd generally seen custom drivers as a way of by-passing native OS audio software and its limitations and thus probably "a good thing".

Finally, my WaveIO board arrived yesterday - many thanks. After a few days to install it properly, I'll report back on how it compares to Doede Douma's USB-to-I2S interface c/w TentClock, a design that was good in its day but is now rather dated. (In any case, the board died on me or, to be more honest, I accidentally fried it.)

[soundcheck]

Originally Posted by Ryelands

You chose to abandon an earlier "must have" route - a low-powered Linux box - and chose instead to modify the Squeezebox with, according to reports on the forums, considerable success. So I'm slightly surprised to see you looking anew at a PC interface before deciding publicly that it has little to offer.

My low power Linux box doesn't sound worse then the Touch via USB btw.

The driving factor to jump on the Logitech train was and still is the IMO best
server and control multi client environment out there. And all that free of charge.
Todays PC based Squeezebox player solutions make it even more flexible.

Unfortunately the SBT project turned out to become a tweaking nightmare as much as my PC journey the years before. Actually it got worse.
Now the network and server had to be tweaked on top of the actual transport/renderer. To be honest. I didn't expect that to happen, when I started the journey.

My current setup sounds really good to me. However. If I'd get access to a solution which performs extremely well without all those upstream tweaks
I'd go for it.

One of the main problem is that the majority of engineers just don't seem to have answers. I'm even wondering if they know the questions?? Probably not.
Many of the engineers out there are PiggyBackers to a certain extent. They just license USB interface solutions. That's why we IMO don't see much progress out there.

In fact they tried to sell for the last couple of years Async USB as the holy grail of PC based audio. Obviously they were wrong - proved by the (ten)thousands of PC tweaking people out there. Those people probably spend countless hours and more money on the tweaks then on the audio interface. The latest insanities are >400$ USB interfaces and audio grade SATA cables.


Then what's the actual poblem? Why does sound any PC setup different?

IMO there is no clear answer yet. For sure there are many factors.

However. I'm positive that we'll see a solution sooner or later.

From what I read WaveIo means another step up. That's good to see.

Originally Posted by Ryelands

Incidentally, I may well have missed the argument but I'm curious why you lay such stress on an interface not needing a "custom" driver. I'd generally seen custom drivers as a way of by-passing native OS audio software and its limitations and thus probably "a good thing".

1. Custom drivers usually lock you up on certain environments
(Try Linux, iOS or soon Android, sometimes Windows only or MAC only,
sometimes they are tight to certain OS revisions.)
Your flexibility goes down the drain.
2. Custom drivers increase cost for the manufacturer (build and maintenance!!)
And guess who's gonna pay the bill.
3. Custom drivers pretty much by-pass the audio interface limitations first of
all.
Spinoffs like 32/384 are nice-to-haves for <0.001% of the users out
there. Though those features seem to work well as sales argument.
4. Custom drivers usually mean much shorter livetime for your device.
Especially if you count on very small niche manufacturers.
5. There's probably more...

A nice example is the new M2Tech Hiface II. It's now "driverless" (custom driver is history). Guess why? From now on supporting HiFace I drivers, will be a pain in ... from a
M2Tech perspective. All those people who bought the Hiface 1 and other OEM devices based on that HF-1 can start praying for long term support.

Cheers

[barrows]

"I'd never buy that interface.

To me the goal really is to hook up any transport to the audio interface and end up with the same sound experience."

Well, I think that you may be waiting an indefinitely long time for that. Transports have always mattered, and probably always will matter. While I agree with your sentiment, I suggest that it is unrealistic. But, the advent of well designed asynchronous interfaces has reduced the impact which the sending computer has on sound quality.
One thing I am confident in, is that current SOTA asynchronous USB interfaces make it pretty easy to achieve digital audio performance which surpasses the performance of playing a physical optical disc, either using a Mac or Linux server without any drivers necessary. Additionally, we have now gotten to a level where 32/384 and DSD playback is possible without additional drivers on both Mac and Linux, and the pace of development is quite high. Considering the youth of high end computer audio playback, I am quite impressed, and do not see myself going back to optical discs ever.

RE: drivers, vs USB class 2 compatible interfaces, it is interesting to note that M2 Tech has now adopted the driverless (I assume XMOS based) approach for their HiFace 2... So much for the idea that a custom driver based solution is inherently superior...

[Wolfsin]

I love scepticism, it gives rise to the very best sorts of "belly laugh" humor. My sides ache and I feel your pain.

As long as the final output is as hard to measure as sound quality there will always be snake-oil salesmen selling "audio quality" sata cables I suspect there will be mass produced, all in one, portable bit-perfect playback devices -- my Roland Edirol amplified through "the wire" into HD-600s sounds pretty good but . . . I would prefer balanced output and 96kHz is limiting and the SSD is limited to 32GB and . . .

Tweaking digtal is a fools errand when bit-perfect i2s with picosecond range jitter is available for a hundred Euros.

[Telstar]

Originally Posted by Wolfsin

Tweaking digtal is a fools errand when bit-perfect i2s with picosecond range jitter is available for a hundred Euros.

computer noise is still a problem until 400Mbit+ optoisolators will be available (i would be happy with 100).

[audiodesign]

XMOS - USB I2S DAC controller

USB - I2S - DAC Controller - YouTube

[triode_al]

Originally Posted by audiodesign

XMOS - USB I2S DAC controller

USB - I2S - DAC Controller - YouTube

A bit short track. Just hear some noise. What is that display from?

[audiodesign]

new video

XMOS USB to I2S and DAC Controller - YouTube

[audiodesign]

Barrows,
probably M2Tech pass to USB Audio Class 2 and forget the custom driver because they have ear the sound of XMOS based interfaces.
In a direct comparison the XMOS is much better than original M2Tech OEM module version 1.0.