One project which I'm thinking about is a box which takes the 2 channels from the empeg and produces 6 channel (FL, FR, BL, BR, Centre and Sub) output. The DSP inside does all the calculations on phase, delay and volume to correct for your off-set position within the car. In other words, it produces a signal that sounds as if you're sitting in the centre of the sound field rather than at one corner (as you are in a car). The box is controlled by the empeg and/or serial connection to a computer running appropriate software.

The software presents a picture of the four seats in the (usual) car and the positions of the speakers. You can adjust the width of the seats (how far between the centre of the seat and the centreline of the car), their forward distance, the forward distance and width of each pair of front and back speakers, and the forward distance of the centre speaker and subwoofer. It might even be possible to do this in three dimensions - height off floor, for instance, to give it an extra bit of realism. The DSPs would then take the sound and do their magic with these parameters.

This is not a "Hall / Stadium / Room" setting, which is simply a filter and an echo component. This is full spatial processing. The empeg could control which position you wanted to favour - driver, passenger, front pair, back pair, all, or whatever - and other bits and bobs like amount of spatial processing done (mix between dry and wet). There may be other things you could do with it via the empeg, but the main control would be via the software. Of course, the protocol would be opened so people could write their own programs to provide alternate interfaces (on alternate platforms).

So, what do people think. Is this a good idea, are there any things you'd like to see in a box like this, and is it something you'd pay money for?

Save the whales. Feed the hungry. Free the mallocs.

Interesting idea. The beauty of it is it doesn't have to be just for the Empeg. It can be for any stereo. (Although serial control would allow you to write a bit of Empeg software to control it if you wanted.)

Have you looked to see if any other commercial products have already been built which do this? For instance, it sounds like the sort of thing BBE might make.

___________
Tony Fabris

Is this a good idea, are there any things you'd like to see in a box like this, and is it something you'd pay money for?

If it worked as well as you describe, it is something I'd pay serious money for.

If I understand what you are saying, this would be a box with a DSP in it and a serial port. You'd plug the serial port into a laptop computer running your software to configure it, then plug the serial port into the empeg to play it. Could there be two serial ports on your box -- one for the laptop and one for the empeg? This would facilitate tweaking.

...a box which takes the 2 channels from the empeg...

But aren't there 4 channels from the empeg?

I don't think I understand where in the audio stream your box would go. Would it be between the empeg and the amps? Between the amps and the speakers?

Can you offer any indication of timeframe? I'm not asking for month/day/hour of release, but are we looking at this year? Next year? I am getting ready to build a new IASCA competition car, and this box could be a very significant aspect of it.

I think Sony has something already on the market very much like you are describing here, but it is seriously expensive (more serious than I'm willing to get -- well into 4 figures USD) so your home-grown version is appealing.

I am quite excited about this.

tanstaafl.



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

Hi,

Forgive my ignorance, but in my world, the empeg already HAS a powerfull DSP.

Why can't that DSP do the tiny delays on each channel?


Lars


Lars

The four channels from the empeg are really two - Left and Right - just split to have different EQ on the front and rear. You're not playing a 4-channel file, so having front/back inputs doesn't really achieve much. I would expect to include a 5 channel parametric equaliser on each output so you could tune the outputs to your speakers. In other words, you wouldn't use the empeg's EQ for speaker-tuning.

I could include another channel of input, switchable (e.g. by a serial command), if it would achieve anything.

The time frame is very open - don't expect a product at all. I'm not a hardware developer and I would have to work through people who are 2000km away so it would be purely a hobby thing for now. This is as good a forum as any to judge what people think of my plans and to get new ideas.

My objective here is to put out a box with parallel goals to the empeg - a (nearly) fully configurable box that does everything you want. Most of the consumer solutions out there AFAICS have a very limited set of options because it's too expensive to give configurability and it's easier to give presets. I want a box that I can tweak and adjust to give me exactly the response I want.

I'm thinking of going up to 8 pairs because my brother has two dash-mounted tweeters that are a different height and forward-distance from his front midranges. Is this taking things too far, or not far enough?

(BTW, the empeg's DSP does not do this at all - it's only an EQ (AFAIK), albeit a highly tweaked one).

Save the whales. Feed the hungry. Free the mallocs.

Paul, this is a similar idea to one that I have been messing around with for a little while. Like a few major car audio companies you should add a microphone input so it will do real time analysis and phase compensation between speakers. Are you going to have the DSP do real time filter coeficient generation? Also, have you determined which DSP you will use? My personal favorite is the Analog Devices SHARC DSP's. I believe they are currently the fastest for FFT's. I think they have a low cost 21060L series with 1Mbit of memory that is only about $10 if I remember correctly. The upper models will run a few hundred dollars but you probably don't need that much memory/features.

Alex Lear

More on this:

Firstly, the box would store all its configuration in a manner similar to the empeg. In other words, you can take the box to any PC running the appropriate software and it will load up the 3D map and other parameters from the box itself. You don't need to take the parameter file around on a separate floppy or computer.

The parametric EQ on each output would be used to provide filtering to match speaker type - i.e. the EQ is used as a low-pass filter for the subwoofers and a high-pass for the tweeters, etc. I'd rather keep this in the EQ than provide a 'low-/high-pass' filter in addition to the EQ (which is quite powerful enough).

However, from what I know of psychoacoustics the ear uses frequency response characteristics to determine the horizontal position of a sound. For instance, sounds from behind you lose some of the high frequency component coming around the lobes of your ear (some people theorise that this is why the ear is shaped the way it is and isn't just a hole in the side of your head). If you hear a sound that's slightly muffled in that particular way you associate it with something behind you. Another aspect of this is that you're used to birdsong and wind noise coming from above you, and animal growls and thuds coming from below, and this may go toward your vertical positioning of a sound (I don't know if this is actually true, but I was told it was by a reliable source).

Would this mean that it would be necessary to 'correct' the signal coming from a speaker behind you for the high-frequency loss from your earlobes? I don't think it would be that significant - after all, we seem to do pretty well in car stereos as it is - but I'll research it. But for user input do you think this would be a useful thing? or would you prefer to set the EQ yourself and know exactly what it was?

If there was some way of getting the MP3 stream straight out of the empeg, then I could also implement MP3 decode in the box itself. This would mean there was only one DAC step (instead of a DAC in the empeg, an ADC in the box, the DSP, and then another DAC in the box for the outputs). Of course, it can't be the RS-232 output from the empeg, because that's got a maximum bandwidth of 115kbps - less than the 'standard CD' encoding of 128kbps. And, of course, we can't use the USB. The only option left in the current configuration would be to have the box talk ethernet to the empeg - doable but it would significantly increase the cost of the device (I imagine).

I think Hugo mentioned (I think) that there's an unused I2S header on the empeg - a digital out, in case you didn't know. Can someone in empeg confirm this, and tell me how practical it would be to get this to an external connector? That would be the raw audio stream, but that's perfectly acceptable (and saves the box having to buy a license from Fraunhofer for a decoder). A feature for the Mark III?

Keep the feedback coming.

Save the whales. Feed the hungry. Free the mallocs.

It would probably be easier to license the use of SRS from srslabs to do this sort of correction.

Calvin

Finished reading the SRS web site, and this would indeed do pretty much everything I want. SRS does the stereo field calculation and corrects for the omnidirectional nature of the microphones and so forth, and Focus realigns speakers not facing toward the listener - for example, in cars - to bring stereo images up to head height.

I'm currently finding out more from them about implementations and licensing issues. If someone else has done the maths then I'll just buy the end product.

I can see this would make tanstaafl's dream.

Save the whales. Feed the hungry. Free the mallocs.

How many channels would you want a sound field DSP to support?
1-2 - one stereo pair
3-4 - two stereo pairs front and back
5-6 - three stereo pairs
5-6 - two stereo pairs, centre and subwoofer
7-8 - three stereo pairs, centre and subwoofer
>8 - in every panel a speaker!

Imagine that you only want to buy one box - don't consider the option of buying two four-channel boxes and running them side-by-side for eight channels.

Save the whales. Feed the hungry. Free the mallocs.

Not completely the same thing, but this:

http://www.tactaudio.com/web-files/RCS 2.0/RCS 20.htm

is by far the best tweak to any super High-end system I've ever heard (and I've heard a few!). Even with systems that were already damn near perfect really benefitted from this device.

Apart from being able to tailor frequency response exactly to a preset, it also corrects for time differences with incredible resolution.

In time, I'm sure this sort of stuff will find its way into cars as well.

There's another unit on the site, the RCS 2.2, that actually has a separate subwoofer output; of course time-aligned separately as well.

(as it so happens, that truly digital amp Millenium is equally incredible, I'm saving up for one...; wouldn't mind a car version of it).


Cas

Or you could look at the professional audio sound reinforcement market. The latest release from sony is a stereo digital input speaker manager. Unfortunately I don't have all the specs but the short review says:
It provides audio processing and loudspeaker management functions. Programmed via a PC, up to 99 units can ber operated from a single PC com port making the {product} well suited for use at large venues requiring complex control.
Audio Technology, V2 I5

OK it doesn't really do 5.1 surround but there are a plethora of products which would do exactly what we want. But they wouldn't be an integral part of the empeg installation.
Unfortunately nothing much can happen until some sort of API comes out.

I'm not complaining or poking fun, and if anyone wants to put anything in their car, let them. I just want an empeg operated device.

____________________
Murray 06000047

Prices are conspicuously missing from the TacT Audio website, so I asked them in an email:

RCS 2.0 DD         $ 2,950.-    (basic version, Digital in/out only)
DA module          $   599.-    (Analog output, 24/96 converter module)
AD module          $   699.-    (Analog input, 24/96 converter module)

RCS 2.2            $ 4,900.-    One configuration only, Digital+analog in, analog out only.

Millennium Mk. II  $10,300.-

TCS                $ 9,990.-

It would be a while before I can drop that kind of money on a home audio system.

Borislav

An update:

On the down side, I haven't heard anything back from SRS Labs past a "Tell us what your application is and we may send you some data" letter (to which, of course, I replied).

On the good side, I've found out that a good friend of mine is an electronic equipment manufacturer, and I'm going to talk to him and his designer friend at some stage about it.

More as it develops.

Save the whales. Feed the hungry. Free the mallocs.