Hey all!
Thought this thread on avsforum might be interesting to you - basically one of the crew at microsoft outlines the new audio processing features built into vista, including drc!
http://www.avsforum.com/avs-vb/showthread.php?t=713073

Very interesting read. I see I might need to start using Windows again soon
Thanks for dropping a note!

I don't really see it as all that useful, the devils in the detail I suppose, but one bonus is that it should stop the spread of crappy 3d and other effects bundled with soundcards.

All of their performance hype is nullified anyway with the protected audio path where every stage in the pipeline adds an encrypt/decrypt stage to protect the user from themselves

Well, playing with DRC in a nice, noob-proof, GUIfied environment sounds great to me. An easy introduction to some really cool technology, I'd say.

I don't understand your second remark, that's a bit too cryptic for me...?

Some information here

Basically Microsoft are building security for audio/video right deep into the kernel. This is to protect the 'content' from the user (you). This is despite the evidence that people still buy DVDs at the moment even without this protection.

A side effect of this is an extremely inefficient audio pipeline - as it needs to be encrypted / decrypted at each dll/function to stop unencrypted data from being snooped. So adding the HRTF functions while playing a blu ray movie, for example, will slow down a lot more than they make out.

I don't go in for Microsoft bashing, or spelling their name with a '$', but there's no way i'm buying an operating system that panders to the entertainment industry. If that's what they want then it's only fair they buy the OS for me.

Wow. Thanks for the link.

From MS's viewpoint ofcourse, this is a feature - and it is, since it's the only way the big entertainment houses will agree to bring their content to pc's. So you're paying for it, much like you would pay for HDMI sockets in a new screen or HT-receiver.

I agree, it's not a nice feature. It may actually sell more MS Vista licenses though - no open source OS can offer this feature.

Soon we'll all be listening and watching obscure creative commons movies and music

(edit - Sorry for going off-topic!!)

seems like a time for another commision (could be European as well) to do something, like start banning MS products sale in Europe should be a good step forward.

I'm trying to decide if this is dynamic range compression, ReplayGain, both, or something else.

Cheers,
David.

It sounds like psychoacoustically driven & slowly acting DRC to me.

Little off topic, but more interesting:

"Any next-generation high definition content will not play in Vista x32 at all. This is a decision that the Media Player folks made because there are just too many ways right now for unsigned kernel mode code [to compromise content protection]. The media companies asked us to do this and said they don't want any of their high definition content to play in Vista x32 at all, because of all of the unsigned malware that runs in kernel mode can get around content protection, so we had to do this."

http://www.apcstart.com/site/dwarne/2006/0...-in-32bit-vista

"unsigned" "malware"?

Well it wouldn't be the first time, need I say Soundcheck?

Interesting stuff. Maybe it's time to start writing a rootkit that lets you access your content in accordance with your rights

apparantly that's not correct - it was misreported. see
http://blogs.technet.com/windowsvista/arch.../24/450081.aspx

I think (at least here in the Netherlands) it doesn't violate your rights... some law states that you're not supposed to circumvent copy-protection (however vague that is!). So the rootkit plan would actually be in violation of the law. IMO, MS is just helping to enforce that law. As I said before, in a way all this is a feature: it allows you to play DRMed crap on your PC. It would be very uncommercial thinking for MS not to provide such a feature...

But you don't have to use DRM crap (and I won't!!). And as far as I understand it, if you are processing unprotected content on your machine, no such encrypting between dll's is going on. So it's all fine with me The DRC stuff is cool! I wonder about this Replaygain-like idea: does that do its work completely during playback (that's what the description suggests)? You'd need to scan a whole file ahead all the time...?

Re: encrypted stream.

The stream will at one point have to be decrypted to be played.

Skillful hackers will simply put either an Upper Filter or (heaven bless me) Lower Filter on the soundcard driver.

BAM! Unencrypted audio data. Granted, it might have been mangled by Windows Mixer but...

Or even really driven pirates will simply make a USB sound device that instead of playing the audio actually writes it into hard disk.

Hmpf. All this DRM crap is punishing the users.

It might not be that easy, because the other parts in the audio chain may refuse to hand the data to your (unsigned) code or (uncertified) USB-Device.

But I agree with you that this will help nothing against piracy. There are always ways around if there is a way to watch/listen to you content.

Didn't realize microsoft had any expertise in audio issues. Quite impressed.

Unless you can get MS to sign you driver, you won't be able to do that.

Unless you use a root kit. It has already been demonstrated that Vista is vulnerable to being exploited by root kits of a variety of types - much like most other operating systems. That's not to mention the virtualization type root kit which would allow almost complete control over the kernel. Until there is hardware enforced DRM (which Vista supports, but is unlikely to show up in consumer machines for a while) there will be a software based way of getting around any scheme that they introduce.

I would rather lose the ability to play HD content than hand control of the computer I own to an arbitrary hardware DRM manufacturer. I just hope that either hardware manufacturers or hackers give me the choice well into the future.

From http://blogs.technet.com/windowsvista/arch.../24/450081.aspx
It looks like MS are trying to convince consumers that this is a performance issue - that there is a technical reason that 64bit processors are in some way needed to play HD content. I'm guessing that very few 32bit platforms will "meet the performance requirements".

It sounds like a scheme to get us to buy new CPUs and motherboards with hardware DRM.

Yeah, I think it is... just as the DVD-A and SACD formats were a trick to get people to buy what was basically an "improved CD-player with DRM built in". That didn't work well, but this might?

The difference is, with SACD the format specified DRM to be in there. With your PC, you still have the choice not to use DRMed stuff.

I believe that it is a level compressor that works on a decent loudness (as opposed to intensity) measure.

Perhaps something related to this set of ideas:

http://www.aes.org/sections/pnw/ppt/loudness/loudtut.ppt



To your first comment, yes, that is my understanding.

There are two forms of this gain, one affects dynamic range, and is 1-pass. This is commonly used for TV apps and other apps that have wide swings in input level but mostly heavily compressed content.

The other is 2-pass in WMP. It assigns a single gain to equalize the peak loudness of tracks. If you turn it on, it will happen during rip, and there is an option somewhere that allows you to recalculate your normalization using the loudness model.

Edited for bad typing. And again for Woodinville being forgetful.

Please note that this is not true for unprotected content.

64 bit CPUs do not include any additional DRM features. The decision to restrict driver signing to 64 bit Windows is purely practical:

MS launched 32 bit Windows in 1995, so its a little late for them to do anything about SAP.

They could implement it in 32 bit Windows too, but it would mean that ALL current 32 bit drivers would no longer work, something that MS simply cannot require.



Good luck getting MS to sign your rootkit.



In x64 mode, this requires you to exploit the system to get your software promoted into kernel mode, which is VERY different then essentially all other OSes.

If you can do that, the root kit is pointless, you might as well just use the exploit directly to go after the media. Such exploits have been shown in the Vista Betas. They've also been very quickly patched. Its not clear at all that this will be possible in the release version of Vista.



Yeah this is how I would do it, but I think you underestimate the challange involved. Cracking the SAP still involves fooling the signed audio driver, which means essentially implementing a software version of some real hardware device on the virtualized machine. If you don't do this right, the SAP will detect what you're doing and refuse to work (probably by simply not loading the driver into kernel mode in the first place).

This is nontrivial to make work, and so far I don't know of any software out there that can do it. I also doubt MS will allow SAP to work on the emulated sound devices in VMWare or VPC.

The MS system is actually pretty well done. My guess is the hacks we see will be based on exploiting third party software. Its probably easier to read the output of whatever decoder powerdvd uses then to actually work through the MS stuff. At least assuming MS doesn't totally botch this by leaving a buffer overflow in the device driver code or something equally trivial.

Personally, I would love to see more discussion about the loudness EQ, room correction, bass management, etc.

Oh, and the post-KS HD-Audio-compliant low-latency interface. Hopefully, this means no more shouting matches between the KS/waveOut/DS/ASIO folks.

It really is amazing to think just how much stuff MS is cramming into the audio stack alone.

Yeah, the whole DRM thing is getting old. I'm quite excited about the room correction stuff. Is there any trace of it in the beta releases (if anyone around here is a tester)? You'd expect this feature needs some testing and user feedback...

low latency? Where did you read this? (I may have just overlooked it..)

So the changes in Vista are (feel free to correct me):
1) Reworked Kmixer (and gave it another name or the like)
2) added some fancy DSPs
3) added protected audio path to protect protected content from the user

1) seem to be a good thing, at least
2) apears useless to me (I guess it's some kind of 'miraculous wonder'-dsp)
3) another thing to bug the user and hand him over to a handfull of huge companies


I would also appreciate some beta-testers experience-reports

Well, if you have a class driver solution in Beta 2, go to the effects tab, have your omni or cardioid (not close-talking) mike handy, and get ready to fool a bit with mic levels.

Then turn on "room correction" (on some speaker setting, not headphones!), and have at.

There is no more Kmixer. Repeat after me, there is no more Kmixer. None. Gone. There is, no longer, any kernel-mode DSP processing to speak of. Resampling is done by one of the DSP's in your point 2). There are several resamplers, of various qualities, for fast vs. clean, as well.

Come on now, you can repeat it, "there is no more Kmixer". Is there something unclear about this statement? You might read EHO's comments in the thread at avforum, maybe that will help somewhat.

Let's see. A high-quality resampler, room correction, loudness equalization, bass management, speaker fill, headphone virtualization, as well as keep the LtRt, and probably something I'm forgetting offhand. I guess "fancy" is in the eyes of the beholder here.

See the many comments above that it is the content PROVIDER who gets to decide if content is protected. You don't have to have protected content unless you insist on acquiring it.

Hard to tell from what you say, since there is no more Kmixer.

It would be helpful if you read the article referred to in the OP here, then you might be better informed about what kinds of DSP are included.

One thing that is not mentioned directly there is that there is also a new, high-quality resampler.


I'll let you explain that further yourself. If you don't buy protected content, you shouldn't get bugged by it.

>>So the changes in Vista are (feel free to correct me):
>>1) Reworked Kmixer (and gave it another name or the like)


>There is no more Kmixer. Repeat after me, there is no more Kmixer. None. Gone. There is, no longer, any kernel-mode DSP
>processing to speak of. Resampling is done by one of the DSP's in your point 2). There are several resamplers, of various
>qualities, for fast vs. clean, as well.

>Come on now, you can repeat it, "there is no more Kmixer". Is there something unclear about this statement? You might read
>EHO's comments in the thread at avforum, maybe that will help somewhat.


There may be no 'Kmixer' but there is still some kind of mixer and controls for each device as well as mixing going on as long as you don't use a dedicated channel (can't remember how they call it) which will also bypass all DSPs.
WMP or WMC or however it is called is not able to do this as well as no other app not specifically designed for vista will be able to do this.
I read the thread in the forums.



Most of those will currently, at non-vista, need not only specific apps but also knowledge and a lot of settings-fiddling. MS is going to implement that usable be the normal 'John Do'. I wish them good look and a wonder, because that must be some sort of automagical wonder-dsp.
Or I am wrong and they implement some more or less geek stuff directly into the OS.
Anyway, I'm sure I will read a lot about those DSPs in future..


Therefore I wrote protected content.




Maybe no Kmixer, but hopefully better mixing and volume controls (really a huge difference to kmixer..)


read it, answered above...





If you want to use protected content your choice in applications will likely be very limited, and if you choose to not use protected content your choice in content will be limited.

Edit: No idea what's wrong with those quotes... sorry, that's just too fucked up to work, for some reason...

There is a gain for each input stream, and then a master gain. The mixing is done in float, and then there is a peak limiter that will "duck" gain when you go over 1 on the output. (No gain in the mixer goes over 1, so that will only happen for strange content that comes in float, or for times when something mixes to over 1. There is no "limiter" operation when the content remains under +- 1.


Well, the resampler gets called automatically. Bass management comes with "average" settings and the ability to change settings and sub polarity. Room correction can be done with a cheap mike or a good one, with better results for the good mic, of course, but good results relative from speaker to speaker with a cheap mike. Loudness EQ comes with settings suitable for most content and the ability to change it.

And so on. You get to play if you want, and you don't have to if you don't know how. How else could one arrange this?

As to settings-fiddling, no you don't need lots of settings-fiddling. Tsk! That's the whole job, make it work w/o infinite tweaking.


Neither do I, I've tried to sort out some stuff here.

Interesting, some guy over at AVSForums has asserted that "room correction via EQ is nonsense", "the room correction crap doesn't work in recievers"... I'd cite it if I knew how to capture the right URL, but I don't, so I'll just say it's near the end of Page 10 of that thread in this vicinity http://www.avsforum.com/avs-vb/showthread....ge=10&pp=30

Sigh...

I wonder.

Correcting first arrival isn't EQ, but how can we get people to realize that there is more to room correction than EQ?

Require people who post on audio forums to have EE degrees?

Or, more specifically, try to assuage people's fears about signal processing in general. I think the two issues go hand in hand, actually. Those who mindlessly fear or ignore DSP usually do not comprehend what goes on in music production or reproduction.

(That said, one can think of a few high-profile examples of DSP gone amok, like iPod equalization, and I think most people think of DSP specifically in those sorts of terms.)

Well, I also enjoyed the guy who wanted to have 'exact bit output' but who also wanted to have Bass Management and Room Correction operating at the same time.

I guess I'm just a tad confused?