Windows 7 Music Server - The First 48 Hours

I installed Win7 on two machines ... one a 64 bit upgrade and the other a 32 bit upgrade. Former went fine. Latter has been a nightmare. <br />

<br />

There is a conflict/bug in my Dell 1210 Broadcom NC 440x 10/100 Integrated Controller drivers. None of the drivers, going as far back as XP and as new as the latest release, work. Has trouble holding an ethernet connection, Auto speed control does not work, etc. My Tech Support guy spent 5 hours on this to no avail. It'll hold a connection if connected directly to the router, but only if the speed/frame settings are precisely tweaked, and won't hold a connection on any connection that has an extra device in between (like a hub).<br />

<br />

Based on this I would only switch to Win7 with the purchase of a new machine.<br />

<br />

Eventually, complete change out of the motherboard, which contains the controller, fixed the issue. I suspect something about the upgrade process caused some firmware on the board to get corrupted, or, reset something deep in the BIOS that the tech support teams don't yet know about.

Have been running Foo + Wasapi on 7RC for 5 months now - no problem so far. I get crashes (still) from Firefox and 7 but that is a 32/64 problem.<br />

<br />

Sounds very decent. Not as good as MPD & linux but ok.<br />

<br />

Cheers<br />

A

Juli@ works fine with Windows 7 x64 final using drivers v1.05.<br />

Actually MME hangs only if latency is large (like 2048 samples).<br />

WASAPI now works in J.River, as long as the buffer is smaller then 1.5 seconds.<br />

I'll try comparing sound quality between WASAPI and ASIO later.

Chris,<br />

My Dell Dimension 8400 running XP Pro, using MM, and equipped with a Lynx AES16, connected to my Benchmark USB with an AES cable, just died. I've repaired/replaced the motherboard and power supply on previous occasions, and am thinking that it's time to buy a replacement. I will need a machine that the kids can do their homework on, so something akin to the Dell Inspiron 530 you've written about might fit the bill. Do you have any further thoughts on Windows based operating systems? I'm particularly interested to know what you hear when you use your Windows 7 system. Any further comments to offer after the frustrating first 48 hours?<br />

Many thanks!

Important question (or is it a dumb one??)... using WASAPI exclusive, is ANY digital out of a PC necessarily bit-perfect??<br />

<br />

For XP I had an M-Audio Audiophile USB via ASIO. Now on Windows 7 via WASAPI exclusive, will this be bit-perfect too??<br />

<br />

And how about the SPDIF out of ANY mainboard out there?<br />

<br />

Can we conclude that getting a bit-perfect out of a PC has just got significantly cheaper? Or in another way, exclusively for digital audio out, why would anyone pay extra for a 'good' soundcard in a Windows 7 platform?<br />

<br />

Am I missing something? thanks.

Hi Neovibe - I wish I could say yes to everything in your post, but right now it isn't that simple :~(<br />

<br />

It's really hard to guarantee bit perfect because everyone's PC is so different and many things can creep up that other people would never think of.

ok, I do understand... some things are so complex than unless we have a low level detailed understanding of it all, then, in general it becomes more of a random phenomenon, a black art... or audiophile <br />

<br />

more to the point now, m-audio usb soundcard (with windows 7 drivers) bit-true or not? what's your opinion?

It depends on the driver, playback software, volume controls, output method (ASIO, WASAPI, etc...), and settings within the playback software. This is one of those things that makes a Mac so appealing to people. I'm using Windows 7 hardcore right now with a couple soundcards and I like it. Mac isn't the only game in town it's just the easier one.

Of course WASAPI is bit perfect in exclusive mode. And Chris, this time it is a plain 100% sure. Haha<br />

<br />

But, let's keep in mind there are a few more or less fuzzy reasons why the output may not be bit perfect afterall :<br />

<br />

1.<br />

The player messes up. Thus, bugs. Not much likely in general.<br />

<br />

2.<br />

You use something from the player that won't allow bit perfectness, such as using the digital volume of the player (Windows volume is not operable anyway).<br />

<br />

3.<br />

You <cite>think</cite> you use Exclusive Mode, but don't afterall.<br />

This may be the most tricky one, and depends on the player. The players I know of which guarantee using Exlusive Mode at all times are Foobar and XXHighEnd. All of the others may switch to Shared Mode (which is not bit perfect) unnoticed because of various reasons (like the wanted sample rate not being supported by the hardware, thus incurring for resampling which Exclusive Mode CAN NOT do).<br />

<br />

4.<br />

The hardware (soundcard) resamples. This is highly unlikely these days, and happened with the older Soundblaster stuff (to 48KHz).<br />

<br />

So yes, your M-Audio will be bit-perfect on W7.<br />

And yes SPDIF out from whatever main board is bit perfect.<br />

<br />

But ... this (again 100% sure) does not imply all will sound equal. The SPDIF out from the main board will sound like sh*t (never mind it is as bit perfect as it can be), as Foobar will sound competely different from XXHighEnd never mind their output measures equal (in the digital domain). I even managed to let sound the two (currently remaining) sound engines from XXHighEnd sound *completely* different, and yet both are bit perfect.<br />

<br />

In the end you can very well conclude that "bit perfect" doesn't guarantee anything for "the best" sound.<br />

What a life ...<br />

<br />

Peter

You are right, but I want to add some small points.<br />

<br />

You said, being Bit-True, does not guarantee perfect sound, that’s right, but without being Bit True, you are far from having perfect sound.<br />

<br />

So for example with iTunes under Windows 7 you can’t get the same Bit-True result when adjusting the correct Sample Rate, as with iTunes under Mac, no way.<br />

<br />

With the actual QuickTime under Windows 7 you have three different output modes and also a “WASAPI” mode, but this is not exclusive WASAPI, so also not Bit True.<br />

<br />

As stated PeterSt, with exclusive WASAPI mode, windows has no control of the volume (lucky it is not), so you can use this behavior the other way round. If you choose for example QuickTime (under Win7 or Vista) and select the “WASAPI” mode you will see when playing back a file, that Windows opens an extra Volume slider named with “QuickTime”, so you clearly see, that this can’t be Bit True, because it goes over the Windows Mixer. Instead if you are running in Exclusive WASAPI Mode J.River Media Center 14 or Foobar2000, you will notice that no additional slider will appear in Windows and that you have no control of the sound with the windows output slider. So this way you have real 1:1 Bit True output with automatically sample rate switch and also with 88.2 kHz sample rate output, that you will not have with the Wave or Direct Sound out in Win7 or Vista.<br />

<br />

Juergen<br />

So a couple of conclusions:<br />

- bit-perfect is a requirement but not sufficient<br />

- "Mac isn't the only game in town it's just the easier one." (Chris)<br />

<br />

It still feels strange to me that 2 bit-perfect signals can sound different. In the end it does not matter, each one should listen and if there's no difference pick the cheapest solution, if there is a difference go for the best you can/want to pay. Problem is from a when you need to make choices without much chance of listening to the several options... then this sort of discussion is indeed helpful (no matter how pointless it may seem to some people).<br />

<br />

But as Chris pointed, Mac is always the easier option... how many threads are there discussing bit-perfect in apple platforms? It's guaranteed out-of-the-box (ok, so then there's the Amarra question...) but it's always more expensive in hardware and much less flexible than PC... time to make a decision....<br />

<br />

thank you all for your clarifications

Judging by the forum posts, WASAPI under Vista and W7 is still pretty shaky. My own experience is that in a gross sense, it works (sound comes out), but there are ticks and pops. This is going out via USB and 1394.<br />

<br />

Juergen, Chris, have any comments? Is Media Monkey or Foobar any better in this regard?

Hi Nicholas - Good timing. WASAPI with my Lynx card works perfect. Via USB is another story. I am actually delivering a specific DAC to J River tomorrow so they can work out some WASAPI USB pop & tick issues.

I am in hurry, because I have to leave soon, will be out of office for two days, but are these clicks and pops with a specific version of MC14? I was running MC14 at the January CES all show days on an USB DAC without any click during the complete show.<br />

<br />

Next week I will be in Far East for another show and will use J.River MC14 on my portable MacBook via BootCamp Windows 7 and exclusive WASAPI, so I will make some testing with the actual version of J.River when I am back in office before I head to Far East.<br />

<br />

Juergen<br />

Hi Juergen - I am using the absolute newest version available in the JRMC forum 14.0.147.

Now I am back in my office and was able to do some newer tests on J.River MC Version 14.0.151<br />

<br />

OS and Hardware: Windows 7 (32-Bit) Home Premium on MacBook 2.1 via BootCamp 3.1.<br />

<br />

J.River output via exclusive WASAPI Mode with the following detailed settings:<br />

Yes on: Open device for exclusive access<br />

No on: Flush device buffers on startup<br />

Yes on: Flush device buffers on pause<br />

No on: Present 24-Bit data in a 32-bit package<br />

<br />

Output set to 24 Bit in the output format field of the DSP studio.<br />

<br />

This test I have done on three different USB DACs:<br />

1. 16 Bit / 48 K, Burr Brown PCM2706 Interface, Isochronous adaptive Mode<br />

2. 24 Bit / 96 K, TI 1020B, CEntrance Code, Isochronous adaptive Mode<br />

3. 24 Bit / 96 K, TI 1020B, with Isochronous asynchronous Mode<br />

<br />

The USB DACs where connected to a root hub of my MacBook (the right hand side jack of the two USB ports, that are on the left side, not the left hand side (this is shared with the keyboard)).<br />

<br />

File Test: J.River MC14 (Version 14.0.151) does play back WAVE, AIFF and FLAC all with Tags and Album Art (even with the Wave File (if stored with the BWF broadcast extension).<br />

<br />

Bit True: It plays 16 Bit and 24 Bit Wave, AIFF and FLAC Files, all with automatic sample rate change and every type 1:1 Bit True (tested with 44.1, 48, 88.2 and 96 K).<br />

<br />

I have absolutely no stuttering, or drops or anything. Everything went perfect.<br />

<br />

Juergen<br />

<br />

PS: For testing the PCM2706, I have set the Bit width in the DSP studio to 16 Bit and limited the test to only 44.1 and 48 K.<br />

Hello Juergen,<br />

<br />

Thanks for the great report and data. I'm running Vista 32 and JR .151, and have gotten good, but imperfect results so far. <br />

<br />

BTW before I forget, what do you think about the dSC White Paper on "bit perfect" from WMP and Vista. I wonder how they're doing it? They don't mention WASAPI as I recall.<br />

<br />

Using WASAPI to bypass the WMP volume control is a big step forward from the audiophile perspective. WASAPI gives a rather different soundstage and timbral balance. With WMP, the soundstage "bulges" forward into the room, and instruments seem rounded, full (in a rather pleasing way) and larger-than-life.<br />

<br />

For example, using WMP, in a Mozart piano concerto, the piano in the center tends to be pushed forward, too large and "deep", timbre-wise, but overall, the sound is really very good, rich, even if probably a little too rich. The variation from reality is definitely euphonic and enjoyable. This is the Bavarian pastry-shop of music.<br />

<br />

With WASAPI, "everything" falls back into its proper place spatially; you see the orchestra "hanging" in a horizontal line in front of you, with images properly sized. There's excellent soundstaging, using the Lavry DA-11, with layers of depth from front to back, and room-width laterally. <br />

<br />

Compared to WMP, with WASAPI the instruments have a somewhat leaner, smaller image, but overall, the sound much more much accurately reflects that in a real concert hall with real instruments. You can just about fool yourself that you're at a live performance. And with female vocalists, the illusion is just about complete. WASAPI is definitely a big improvement. The Lavry DA-11 brings out the chromatic radiance of the piano, and one has the distinct impression that it is a primarily wooden, albeit steel-stringed instrument, rather than a glassy or metallic one. Violins sound like the fragile contraptions that they are, and the sounds emitted have a totally natural, nourishing, heart-softening raspy, but pleasant buzz to them, sort of like the sounds of insects in the forest.<br />

<br />

WMP via 1394 is about 6 dB louder than WASAPI even though all enhancements are disabled.<br />

<br />

But there are still (somtimes) annoying ticks and gaps using WASAPI via J River on USB or 1394. It seems to go in phases, with long periods of time without any problems.<br />

<br />

Moving over to USB, on my Toshiba Qosmio machine, there are 4 USB and 1 USB 2 controllers showing up in Device Manager, with 4 USB ports. The one used by the DAC shows having 45% of the available bandwidth reserved for it, which should be plenty. So apparently the Lavry is a USB, not USB 2 device.<br />

<br />

USB is about 3 dB louder than WASAPI, and the sound is brighter, more detailed, relatively unnatural compared to WASAPI or S/PDIF via WMP. The soundstage is flattened front to back, instruments and voices are wider, grainy, and massed strings are piercing. Overall WASAPI via USB sounds about the same as WMP via USB.<br />

<br />

With USB and WASAPI, one has to set the number of channels and bit rate, or JR gets confused. With 1394, the settings can be left at "determined by source".

Nicholas I am a little bit short in time, because I am on business vacation, but some small comments from my side:<br />

<br />

Something is a bit strange for me, because if you set your software to be Bit True (either JRiver MC14, or Foobar 2000), it has the same level whether you use the WASAPI driver or the ASIO driver, even Direct Sound or Wave have very similar levels (they differ only in the 24th bit).<br />

<br />

And additionally, I can use my SPDIF Out, or AES/EBU Out, or USB out or Firewire Out, into one of my different DACs, I have absolutely the same level with each interface. Why should I have different levels, because for example – 20 dBFS is the same digital value on every interface?<br />

<br />

The dCS white paper is in my opinion too much simplified, so I would not use it as reference. The dCS DACs are great, without doubt, but this white paper about computer playback is far from that.<br />

<br />

Juergen<br />

Hi,<br />

<br />

I would have said the same as Juergen, including my opinion about this paper (there is another thread in here about it, including my "ideas" about it). But about the dB values I thought this could be the explanation :<br />

<br />

It is clear to me, no matter all is bit perfect, the one player (or audio engine in there) can easily show a louder (perceived ?) SPL than the other, because of more harshness and the like. So, for me (and my community the same) it is often about "hey, I can play much louder now !", and it is supposed to be a virtue (less harshness etc.). Now, the only additional thing it takes, is that this "harshness" can be captured by a (measuring) microphone, and it can show the additional SPL coming from it.<br />

Btw, I once had an XXHE version which clearly showed 20dB more SPL in the sub low region. Still bit perfect.<br />

<br />

So the question to Nicholas : were you measuring this by means of a radioshack (etc.) SPL meter ?<br />

Peter

Hi Peter,<br />

<br />

Thanks for the comments. Sure, my observations were based on subjective estimates and objective methods, which, for once, agree closely.<br />

<br />

With a trusty RS SPL meter, with pink noise as well as 315 Hz and 1000 Hz tones, the results at the listening position are:<br />

<br />

Windows Media Player<br />

88 db, 1394 to SPDIF to DAC <br />

82 dB, USB to DAC <br />

82 dB, TOSLINK to DAC<br />

<br />

JR River (WASAPI and DS modes were identical)<br />

82 dB, 1394 to SPDIF to DAC<br />

86 dB, USB to DAC<br />

86 dB, TOSLINK to DAC <br />

<br />

So the initial subjective observation that the WASAPI/DS from JR was 6 dB down from WMP was about right. Why these numbers vary at all between players and modes is probably obscure. Presumably the DSP within the Windows audio stack is part of the answer, and the entry-level 1394 to S/PDIF is another part.<br />

<br />

Getting digital correctly isn't easy.

Hi Juergen:<br />

<br />

Does business vacation mean that you're making money and having a good time? In case you're in Hawaii, come on over.<br />

<br />

With some additional listening, I'm getting very good results WASAPI via (glass) TOSLINK (another Wireworld cable) to the Lavry... no ticks or pauses. TOSLINK still sounds way better than USB.<br />

<br />

Thanks for the comments on the dCS paper; it we can get the story behind the measurements, they will probably make a lot more sense. Channels are being opened...<br />

<br />

Bon voyage...<br />

<br />

Nick<br />

<br />

Hi Nicholas,<br />

<br />

We better forget about my earlier remarks; something else must be going on. 4dB for pink noise is a lot, and nothing like I suggested - which would be about certain frequencies and/or harmonics combinations.<br />

<br />

This can hardly be all bit-true.<br />

But maybe it is an idea to call it operator error ? haha. Measuring properly with an SPL meter isn't all that easy, and while you must keep the distance the same (easy) there also shouldn't be something else in a different place (YOU !) because of reflections. And maybe you didn't see it, but your last list isn't 100% consistent with your post earlier.<br />

<br />

Anyway ... I don't recall performing (SPL) tests like this to relatively judge a player or playback means, and maybe you shouldn't do it either. Maybe it is better to use the known theories, like WASAPI being bit-true if used in Exlcusive Mode. Or read Juergens posts. Just use it, and if you like the sound, keep on using it (btw, your description of the sound of WASAPI is creapily close to what I would say, compared to something like DS).<br />

<br />

Also to keep in mind : WASAPI doesn't contain or produce ticks and the like. Oh, it sure may differ from player to player how easy it is to incur for them, but try to keep in mind it's always your PC. So, never mind with e.g. ASIO everything is allright, it still is your PC.<br />

<br />

<cite>Getting digital correctly isn't easy.</cite><br />

<br />

Not yet. But we are learning.<br />

Peter

Oh, oh, Nicholas, this sounds wired.<br />

<br />

I would recommend, you leave WMP at side (at least for this moment, or forever, because I can't play bit true), in concentrate on JR MC14. Go through the settings, that Chris has posted at his J.River post, and this will guide to you good sound. And if you are happy with Toslink, that's fine, go for it and take it.<br />

<br />

Good luck.<br />

<br />

Juergen<br />

<br />

PS: I am sorry, I am not in Hawaii and can't visit you, right now I am in Asia, and yes, business vacation means making money and also having some good time, but I must tell you, I love my job a lot.

I agree that WASAPI bit-perfect through J River is definitely the way to go, for our situation. The only problem right now is that the 1394 output seems to have some ticks and pops; however, the 1394/SPDIF interface may well be the problem. <br />

<br />

USB and TOSLINK SPDIF are at least tick-free, and while the USB audio is not acceptable, the TOSLINK with a glass fiber is suprisingingly good. 1394 is better, but at the moment is a bit problematic with the tick artifacts. Seems like hardly anyone has software that will run on typical Windows (meaning HP, Acer, Asus, Dell) laptop 1394 chipsets.<br />

<br />

Have fun on your trip! Nice time of year to go to Thailand; probably still a bit chilly in Japan and northern China. Keep in mind that it's only 7-8 hours from Tokyo to Honolulu; about 10 hours from Taipei.

Nicholas:<br />

You say "USB and TOSLINK SPDIF are at least tick-free, and while the USB audio is not acceptable, the TOSLINK with a glass fiber is suprisingingly good."<br />

<br />

It appears you are referring to USB going out straight to a DAC. What about going through a converter such as the M2Tech Hiface or one of the others around? Is the only reason you say USB is unacceptable is because it will not playback higher resolution than 48 KHz native or is there more to it? Curious to hear what you have to say.<br />

<br />

Chuck