Ok, for those of you who are familiar with the EMPEG MKII architecture and software, how possible would it be to implement a variable delay on a per channel basis (L, R) in the EMPEG? (i.e., somehow pass the L and R audio signals through independent variable size buffers before being DACed - or alter the size of some software buffers that may already exist in the EMPEG, etc). The delay required would be on the order of 10 ms or less.... I could come up with some real sets of numbers if this was actually possible.

Why am I asking this? This is essentially the same thing many high end car audio head units do to compensate for your seating position in the car which they typically refer to as time alignment (although wat I described would just allow for compensation in the L/R plane, while some of these head units allow more complex permutations like F/R adjustment, echo effects for the rear, etc).

If you are not familar with time alignment or it's purpose.... the simple explanation is that it can be used to equalize the arrival time of the L and R channels for a given seating position in the car. For example, if you are in the driver's seat of a left hand drive car, then you are generally much closer to the left speaker(s) than the right speaker(s).... where as the ideal for a stereo recording is to be equi-distant between the left and right speakers. The time alignment just compensates by delaying the closer speaker. Speaker location can reduce the degree of this problem (such as kickpanel installations), but usually not totally eliminate it. The benefit of doing this is a potentially better soundstage and imaging (for the one seating position it is optimized for). I have a somewhat high-end system (with three way components in kickpanels) I use for IASCA competition, and even though the sound stage and imaging are pretty good without time alignment, the sound stage is notably better with time alignment. IASCA judges generally give the system much better subjective scores (which are largely based on perceived sounstage and imaging) with time alignment.

I know there are processors that can do this, or I can make a simple ADC/delay/DAC black box. In fact, I am using a Sony ES CD player to do this at the moment, but that is the only reason the Sony is still in the car for now. It would be nice if the EMPEG could do this since it could be implemented before the audio is DACed negating the extra AD/DA conversions... so I thought I would ask the experts.

Also, I thought that there might be at least a few others that would be interested as well if it were possible.

AudioControl's website has a brief technical paper on the problems with stereo in the car environment, if you want a better explanation of the problem.

http://www.audiocontrol.com/techpapers/tech1008.pdf

As they note, time alignment isn't the ideal solution, but it does give pretty good results for one listening position at a time, and it is relatively simple to implement (compared to adding a center channel, etc).

Whatever reaction you may have to this, thanks for reading through it :).

Mike

I would like DTA as well.

Taym
_________________________
MK II BLUE/RED 12GB #923

I am using a Sony ES CD player to do this at the moment, but that is the only reason the Sony is still in the car for now. It would be nice if the EMPEG could do this since it could be implemented before the audio is DACed negating the extra AD/DA conversions... so I thought I would ask the experts.

The only problem is... IASCA will not allow you to use your empeg as the source for the competition audio. I have discussed and argued this with them at quite some length, and have reluctantly agreed with their reasoning. In their competitions, you must use their CD, provided by their judges. They wish to eliminate the possibility of anybody gaining an unfair advantage by remastering the original tracks, the object of the competition being to determine who has the best sound reproduction equipment in his automobile, not who has the best sound editing studio at home.

So unless you are planning to use a CD player into the Aux. Inputs of the empeg, having TD capability will be of no benefit in an IASCA competition. And if you are doing that, well, you already have the TD capability in your existing CD player.

Now, for just plain listening, sure, I'd like TD in the empeg... but it won't help in IASCA competition.

What class do you compete in? What's your system like?

tanstaafl.



"There Ain't No Such Thing As A Free Lunch"

Hey tanstaafl,

Yes, I am all too aware of IASCA's current rules and how they basically rule out anything but CD players, etc. I had some email exchanges with Tom Walker about this in the past as well... but to no avail. I even offered to bring an empty EMPEG and a laptop to events and rip the judge's CD under his/her supervision, but even that was not sufficient to allow me to compete with the EMPEG. I guess they really just don't want to go down that road right now.

As for my system, there is a JPG attached which shows what I have now... very simple. In about two weeks, it will be even simpler as the Sony will be removed from the system. You can go to my website (link below) to see highlights of what I had last season.

http://www.mrmunsell.com/audio/CRX_Audio/crx_audio.html

I actually competed pretty seriously last year (it was my first year) just to see how far I could go.... and I made it to the finals, which was a good learning experience (translation: I didn't do as good as I expected :). Although I only got 9th in my class, I did rank pretty high in the subjective sound quality area (I'll have to check, but I think I was at least 3rd in that area). It was all the extra creativity stuff that really hurt my score - I got only one point there (even with the stealth steering wheel controls, the motorized OEM delete panel/relocated CD display, tweeters mounted in the AC vents, etc, etc).

Anyway, I have now decided to just design my system the way I want (one head unit now that i have the EMPEG tuner module!), and then if possible I'll make some sort of provisions to temporarily throw in a CD player just to enter IASCA events. If I can't make that work, then I just won't compete for now... at least not until head units like the EMPEG are allowed.

I now compete in the Street Amateur 1-600 Watt class. I am glad they added the street category.... so one doesn't have to go overboard with linear actuators, gels coats, etc just to be competitive at the major shows. My primary concern is sound quality, while trying to keep the design simple and the install clean.

BTW, I think we exchanged some emails back when I first got my EMPEG earlier this year.... I believe I was originally asking about some issues I was having when feeding the CD player into the EMPEG or something along those lines. You compete in SQ in Alaska, and have a Taurus right?

Many high end head units have some variation of TA features, and that is the only thing I think I will miss once I yank the CD player. So I thought it would cool if that feature could be hacked into the EMPEG (if it weren't a major ordeal, etc). I am still learning Linux and know little about how the EMPEG actually works.... which is why I wanted to ask the experts if it might at least be theoretically possible.

If so, that would be enough incentive for me to make a side project out of learning about what is really going on in my EMPEG and figure out how this might be done. I know there is steep learning curve ahead for me if I went that route, but I'd be willing to at least try.... that is unless someone already knowlegable enough decided to do this or already knows how to do this and wants to share the details (which would be excellent!).

But if this is just not possible at all for whatever reason (which I realize is a strong probablity), then I'll just use my near term 'spare time' on designing and making a simple hardware box to do the same thing - which would actually be easier for me since I have more experience in hardware and have a good idea where to start, etc. I just don't like the idea of the extra AD/DA process.... but if that is my only option, I feel the benefits of TA outweigh the degradation from the extra processing.

Thanks!

A project like this would have to be a kernel hack unless the empeg guys do it since we don't have their player source. I don't know much about linux kernel internals, but it seems like this should be possible. You may want to have a look at the volume adjustment (voladj) kernel hack floating around. It seems like this could be done in a similar manner.

-Mike

A project like this would have to be a kernel hack unless the empeg guys do it since we don't have their player source. I don't know much about linux kernel internals, but it seems like this should be possible.

I too feel like it should be possible. An alternative to a kernel hack might be a library that sits between the player and the standard C library and changes data going to the audio device. You can use the partially linked player binary to achieve this. Doing it this way won't be necessarily easier, just more obvious for somebody with UNIX programming experience but no kernel hacking experience.

Borislav

I have considered using this approach to add Ogg Vorbis support once a fixed point decoder is available and the release of v2.0. When v2.0 comes out it will supports wave files. A libc open() and read() wrapper might make it possible to filter other audio types into the player already decoded.

From the perspective of the folks that have already talked with IASCA, if they were to get more requests to allow the EMPEG (in competition), do you feel they would change IASCA's position?
Just wondering if I should waste my breath pushing the issue.

Thanks, John K

Riocar 40Gig #622

An alternative to a kernel hack might be a library that sits between the player and the standard C library and changes data going to the audio device. You can use the partially linked player binary to achieve this.

Very, very interesting. Could some sort of plug-in idea be build around that concept? If it is possible to use this idea to get into the IO path for audio and file operations (how about video also?) there are a ton of cool things that could be done with that.

-Mike

DTA of left/right is possible and on the wishlist. We can't do all 4 channels though, just left/right.

Hugo

Wouldn't someone that is THIS critical of audio, (i.e. not just an audiophile like most of us but an elitist audiophile), NOT be listening to mp3's anyway? Because of their "inherently flawed' nature?

Mark I / #102 / 48g / Smoke / Custom Fascia
Best PC/Audio Investment Ever Made

Could some sort of plug-in idea be build around that concept?

I guess so. What does the typical API for such plug-ins look like?

Borislav

I have considered using this approach to add Ogg Vorbis support once a fixed point decoder is available and the release of v2.0. When v2.0 comes out it will supports wave files. A libc open() and read() wrapper might make it possible to filter other audio types into the player already decoded.

Yeah, I've been thinking along similar lines with FLAC. One annoying issue is that the player will be caching uncompressed audio, which means that such a decoder won't be very useful in the car.

Borislav

do you feel they would change IASCA's position?

In a word.... No.

Not unless there could be some way of verifying that the .wav file played on the empeg (forget about even thinking about MP3 files!) was byte for byte identical with the .wav file on the original CD. This verification would have to be both fast and unhackable.

In other words, it isn't going to happen anytime soon.

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

What does the typical API for such plug-ins look like?

Maybe something like a LADSPA host? link 1 link 2 That would give a bunch of existing plugins to play with.

-Mike

DarkStorm:

Ouch.... now I am elitist audiophile :). I honestly don't think I am any more of an audiophile than many people on this board..... or at least hopefully I haven't gone THAT FAR overboard with my hobby :). I mean afterall, I don't think many elitist audiophiles would consider 'high-end car audio' a serious phrase anyway. Audio reproduction in the car environment introduces so many problems of its own. The difference between listening to a well recorded non-compressed track in a well balanced high-end car system and a well balanced high-end living room system are typically pretty immense. I would say the difference between playing back a CD and a high rate MP3 codec-ed version of the same CD in either environment are subjectivey much milder, when any real differences can be detected at all (especially in the case of the car audio listening environment).

Seriously though, I do see your point. Along those lines, I usually get mixed reactions to the EMPEG from fellow competitors at IASCA shows. Ranging from complete envy to complete incomprehension (why would I choose to listen to anything that has been through the MP3 codec process). Generally though, the latter reaction is much less common.

I am an EE practicing in the field of MPEG compression, so I am fairly familiar with the artifacts MPEG can introduce to audio and video. And yes, the results can be ugly when the technology is pushed to or past it's limits - and yes that makes me cringe. But in the same breath, I would say that the compression can be pretty transparent at relatively high rates.

But even if the MP3 artifacts are more evident (like at lower rates in joint stereo), the benefits of TA are not lost. TA can still be useful to make the best of these MP3s at playback in a car environment.

The other response to this is that soon the EMPEG won't only be an MP3 player.... one could choose to only playback WAVs, and the EMPEG becomes a very viable high-end head unit even for the elitist audiophile who despises MP3. With all processing done in the digital domain, good DACs, etc - it is like having a high-end CD player that can magically archive and play back about 100 CDs (if you had the 60 GB model).... all that with a much nicer and easier to navigate user interface than you get from typical head unit/CD changer combos.

Well the responses to this post are encouraging.... this sounds like it may be more possible than I was actually expecting, glad I asked you guys.

Hugo: I am glad this was already on the wishlist. I had expected the F/R TA wouldn't be possible from the little I know about the inner workings of the EMPEG (i.e., there is no notion of F/R until the two channel stream gets processed by the Phillips DSP, right?).

Can you give us any idea if this is somewhere on SonicBLUE's software development map? I am only asking because I can imagine there are all sorts of things that are on the wishlist that may never be implemented by SonicBLUE for various reasons.

I think I phrased it wrong.
Perhaps I should have said purist audiophile.

I didn't mean to slam anyone. It just seems to me that the difference in arrival times at the ear in an automobiles environment would be completely undiscernible, at least to me.

Then again my hearing isn't the same as it was 15 years ago in my teens.

Steve

Mark I / #102 / 48g / Smoke / Custom Fascia
Best PC/Audio Investment Ever Made

As tanstsafl said, another person bugging them about supporting non-CD head units probably won't make any difference. Now if the whole EMPEG community decided to start competing and bugging IASCA about this, that would be a different matter :). The conclusion in the last email I received from IASCA about this said:

"Agreed that many of your arguments are very valid. However, for this year 2001 you will need a CD player to compete ...... The answer in future years may be different but for now the CD itself is our best control. "

I got the impression that they are not on the verge of a decision or anything, just avoiding the whole issue for now.

I have a feeling that even if we could find a way to prove a ripped WAV on the EMPEG were the same as the original CD, it still wouldn't be sufficient. Currently, you can't even use a copy of the official IASCA that you purchased from IASCA (i.e., they made the copy, not you, it has their silkscreening, and it's not like they are on CD-RWs or anything either). There really isn't any provision in the rules to even allow you to try to prove your copy of the CD is the same as theirs so you can use it instead, they just insist you use the judges copy.

I am starting to think it will require a significant change in their philosophy before allowance for the EMPEG would ever actually be implemented. I can understand their stance to some degree.... but unfortuntaly it doesn't allow for any source unit that doesn't directly play some sort of physical media they provide - which currently they limit to CD. There are no provisions to even allow you to compete with MD, DAT, etc as far as I know - not that most people would want to use these and they aren't exactly mainstream either..... but even if you did want to use one of these formats for competition, I don't think they wouldn't let you. In the case of DAT, you can make an exact copy and prove it's an exact copy......

So I am not holding my breath anymore while waiting for this to happen - I am sure it eventually will, but it seems to me they won't consider revamping the rules until using such players as a primary head unit is more mainstream.

DarkStrorm: No worries.... I don't feel that I was slammed.... just that maybe I needed to explain myself as I don't want to come off as an elitist anything. But obviously, that wasn't what you meant. I would admit that you did peg me correctly as being an audio purist though - however, I do accept that some non ideal things have to be done to my audio for various reasons (EQ to compensate for my interior transfer function, TD to compensate for speaker location, MP3 to get my entire music collection onto something like the EMPEG, etc).

I don't know if you ever had a chance to play with time alignment or listen to someone else's system with it on vs off, but it really can make a large difference (at least in a car, where speaker location is usually so far from ideal). I know a few ms to one channel or the other seems insignificant. But the difference it yields usually isn't like one of those subtle differences that people try to explain when they switch from oxygen free copper speaker cable to unobtainium plated silver cables or something... it is pretty prominant. Most of my non-audiophile passengers at least notice that something significant changed when I toggle time alignment on and off (I can optimize for driver or passenger seat with the Sony head), though some may not know exactly what is different or if it is better or worse :).

If you hear a sound in a room and can tell where it came from by the sound alone, I think that is all the hearing you need to notice the difference time alignment correction can potentially make if you wanted to listen for it (the difference in L/R arrival times plays a similar role in both).

For example, take my system from last year, which needed more time correction than my current one. I had speakers in the stock lower door locations and tweeters in the dash. There are some tracks on the IASCA CD that IASCA documents where the various components of the sound are supposed to be perceived on the soundstage both to help you tune your system and as a basis for judging. The easiest example - most of the speaking voices to explain tracks are supposed to appear to eminate from somewhere near ear level in the center of the soundstage (center of your car), and ideally the perceived soundstage depth would be out on your hood somewhere. For my system without time alignemnt, the speaking voices appeared to come from my instrument cluster directly in front of me, and was not really distinct. With time alignment dailed in, the voice appeared to come from a more distinct point that was a half foot or so above the center of my dash - still not ideal, but better. Not only that, but playing with the timing can result in better imaging. In many stock stereos, you can easily tell that the example voice above is being reproduced by two or more speakers on either side of the car... and if things are too bad maybe you really can't perceive a soundstage at all from them, you just hear the two speakers no matter how hard you try, it's just too far off to fool your brain. But once you get the timing better by whatever means (speaker location, time alignment, etc) the voice coming from the two speakers starts to sound like one, it becomes harder to pinpoint where the speakers are, and it gets easier to pinpoint where the voice is supposed to be coming from (and it starts to sound more like the example voice is coming from a single distinct point).

Of course, getting everything really good can seemingly require endless tweaking of components, mounting locations and orientations, etc (that's what competing in IASCA is all about) - I have generally found that whatever I have put in and however it sounds, time alignment can usually at least make it at least a little better in the areas of soundstage and imaging (at least for one seating position at a time - which is the main drawback of TA).

Hi.

So I am not holding my breath anymore while waiting for this to happen - I am sure it eventually will, but it seems to me they won't consider revamping the rules until using such players as a primary head unit is more mainstream.

Not that I am interested in IASCA contests, but isn't the cat biting its tail here? Non-CD units wont become mainstream (in IASCA contests), because (most) contestants won't use those as their main head units, because they are basically ruled out. Right?

cu,
sven

proud MkII owner (12GB blue/green/smoked, was #080000113 is #090001010)

sven:

Yeah, pretty much. But I guess I meant when they became more mainstream with the general market. Meanwhile IASCA competitors would have to maintain a CD player in their system if they want to compete, etc.

But now that you mentioned it and make me think through this a little more, maybe the harddrve-only head-unit issue won't be forced on IASCA in the near term future anyway (so now I 'm really glad I am not holding my breath). With the large companies like Sony/Kenwood/etc moving towards players that can play music off CD and hard drives/memory/etc and other like PhatNoise and Rockford Fosgate making changer-like-devices to play music off hard-drives/memory (and SonicBLUE out of the aftermarket car audio game).... it seems reasonable that it will be these mixed CD/hard-drive/memory implementations that will become more popular in the market.

....Sigh....

Regardless, I love my EMPEG and it will remain the center of of my car audio system.... it remains to be the best audio component I have ever purchased by far (and I have purchased my fair share :). And I doubt the things the big manufacturers are going to offer will sway me in the slightest, especially from what I have seen of their intial offerings.

So now I just have to accept that I may always have to throw in a CD player when I decide to enter any IASCA shows..... ok done.

Ahhh.... something just now occurred to me that was probably already obvious to most of you guys. Unless the DSP can do the TA, the TA is only going to be effective for anything that can be processed by the ARM (i.e. files played off the hard drive, but not the aux input and maybe not the tuner output either). Is that right?

Either way, it would still be a cool feature.....

Correct. No ETA though, I'm afraid.

Hugo

I'd sort of fogotten about this idea despite having considered it many years ago. Now, wanting to sync my iTunes library meaning i need to keep translating AAC files to mp3, I'm wondering if it might make sense to embed a AAC decoder and have an open() of an AAC just background decode to a tmp file, and then remap I/O to it; close() could discard it.

Just use a pipe rather than a temp file.

I assume, then, that the player doesn't mmap the songs ever. Which is probably true.

It doesn't, but it does seek around in the file.

temp file it is.

Of course, you'd have to ensure that some partition remains mounted rw.

48mb unit, i certainly wouldn't feel bad about making a memory-resident filesystem to do so, to an extent.

tho, continuing to brainstorm: were there posix extended attribute support, i'd shove an mp3 into an xattr and link a player binary which knew to use that in place of the main data.