I've just discovered this on Doom9 forum...



Main Aud-X Features
Dolby Digital compatibility. Designed to work with stand alone home movie theater amplifiers (AC3 in/out compatibility)

Full scalability. Aud-X track can be played on any speaker and sound card set-up. You can use stereo speakers (by switching the DirectShow decoder to stereo mode), use your PC 5.1 speakers set, or hook your PC to the home movie theater. If one day you will upgrade your sound system – it will cost just to change the decoder setting to enjoy surround sound,

Mp3 backwards compatibility for older software and hardware. On machines without Aud-X decoder installed, users will still be able to enjoy stereo or mono mp3 sound (depending on the encoded Aud-X quality option),

AVI compatible. This function makes Aud-X perfect companion for Divx/Xvid encoded movies. It must be emphasized that Aud-X works good at 128 kbps (traditional mp3 bitrate for movie encoding),

Avi, Ogg, Matroska, Mpeg containers compatibility,

Complete solution. Aud-X comes with a complete set of tools for movie encoding and playback. We also support users with „How to” guides at www.aud-x.com

Wide range of encoding bitrates. Aud-X can create streams ranging from 80 to 192 kbits. You can choose preferred quality accordingly to your requirements

80 kbps - world record breaking compression bitrate for 5.1 sound. Such a low surround sound bitrate allows for successful multichanel Internet audio streaming,

Easy to use interface. Designed for non experienced PC users not, only multimedia professionals and software freaks,

Web site support. FAQ’s and HOW_TO’s at www.aud-x.com

High sound quality. Tests show that Aud-X outperforms other multichannel encoding technologies at comparable bitrates, I'm curious about that...

Built-in Pseudo Surround module. You can choose to listen to your old mp3 stereo records with Aud-X DirectShow decoder. Stereo sound will be converted on the fly to 5.1 format with sophisticated algorithm that adds new spatial feeling to your tunes / movies.

http://www.aud-x.com

some doc

I'm too tired to comment on this, but I will give ONE quote from their white paper:



Needless to say, I don't have high hopes for this tech.

Why don't they just download the source code to Ogg Vorbis and have a developer work on implementing multichannel coupling and optimizing some codebooks? This is exactly what I mean instead they go out and invent something entirely different. Sheesh... I suppose logically it makes sense though. I also noticed that Theora isn't mentioned at all on the Doom9 forums.

Tried the codec, compressed the audio, compared it to HE-AAC Streaming, did not like at all the metallic-like artifacts, uninstalled the codec and kept working with HE-AAC.
Fun fact: the artifacts sounded alike in 128 and 192 kbps (what´s up with that?).
Plus, I keep getting channel crossing even with speakers set to 5.1.

Their encoder sounds like Fhg mp3 surround (I didn't test ACM version, only frontend)...The good point is that their decoder is able also to decode fhg mp3 surround stuff. So, we can mux this kind of stream into AVI/OGM/MKV/MP4 now...


@Garf: sligthly OT but how can we perform some ABX tests with multichannel streams ? Is there somewhere such public tests ?

Hi All,
I am the one responsible for the codec concept.
---
First of all, thank you for comments, and especially those sceptical ones
I will try to write some more about the technology and maybe I will convince some
of you to give it a try
---
Yes. There is a channel cross talk, similarly as in mp3surround. Although we tried to minimize it in terms of its influence on sound direction perception (i.e. we tried to minimize phase errors).
The general concept is that all of the channels are merged to stereo or mono and then (as in mp3surround)
based on some auxiliary information (of limited bitrate) the complete panorama is reconstructed.
---
IMO our 128 kbps is significantly better than Fhg's 192 kbps, but I am not objective here, so everyone would have to judge.
The codec is dedicated to movies, thus testing it on some artifical sounds, such as sinusoids, square, etc. will not produce satysfying results.
---
I think that it would be nice to find out users' opionion with regard to the following questions:
1) Is it better than AC3 128 kbps (for our STDQ 128 kbps)
2) Is it better (STDQ) than mp3surround for 192 kbps?
2) Is better our 192 kbps than mp3surround 192 kbps and AC3 192 kbps?
3) Is it worth to use our 128 kbps 5.1 than 128 kbps mp3 stereo (considering all cons and pros)?
-----------------------
Ofcoruse to test this in reliable way some subjective tests would be required.
Alghogh users opionios could be also interesting.
-----------------------
Why not developing AAC or other concepts?
Well, we wanted the codec to be compativle with the AVI container.
On the other hand, encoding each channel separately with limited bitrate
(to achieve 128 kbps it would be 21 kbps per each channel), instead of merging all the channels also produces artifacts, related to low bitrate compression. Thus it was just a matter of choice...

foobar2000 can do multichannel ABX

Since your baseline codec is MP3, beating MP3 Surround in a test should be considered a very interesting result, I think. (Unfortunately I don't think your testing methodology is valid, and you yourself must know this given the quote above)

I don't think the tech in MP3 Surround was made public but sensibility says that it's very likely to be FhG/CT's proposed method for Spatial Audio Coding (but applied to MP3 instead of HE-AAC). You can see the significance in beating this. (SAC is going to be based on an improved method, but still)

Did you write the MP3 encoder yourself?

The mp3 encoder is Lame.

Regarding this tech, can we have decoding specs please? You probably know that it is unlikely to be successfull without at least decoding specs available.

Yes. The mp3 encoder is LAME, as explained on our web site.
(www.aud-x.com)

So the sound quality is as for 128 kbps LAME 39.6.1 (if I remember correctly) stereo (in terms of the base sound channel quality. Please refer to the whitepaper).
--
Channel cross talk errors
are related to the 5.1 panorama reconstructing methodology, specific for Aud-X.
---------------
In fact this is not the question wether LAME is better than Fhgs base mp3 encoder,
because both for 128 kbps produce good results, in comparison to the channel crosstalk errors (i.e. those errors are bigger problem for our codec, and for mp3surround).
To summarize, problems of mp3surround and Aud-X are channel crosstalk,
not the base signal mp3 encoding quality (IMO).

I forgot to say. IMO
we produced much better results in terms of channel crosstalk errors,
in comparison to Fhg

Like a Prologic decoder... (?)


Tested with a movie audio track.

AC3 is a 10+ year-old codec, to use it as a base for comparisson would be like saying that H264 is good cause it´s better than Cinepak Codec. Compare it with the super-codecs like HE-AAC.

Considering that is pretty much matrix-decoded, then what´s the difference with 128kpbs Stereo MP3 from 2-ch matrix-encoded sources?

Have you guys analyzed Vorbis method for dealing with crosstalk?



Why? I am not a huge proponent of any more containers format's we have enough as there are already, but in this case if I was encoding video I would still prefer Mastroska or even Ogg. AVI is annoying and PITA. I understand what you are trying to do, but still.

Here is how I think it should work.

Ogg -> Theora/Vorbis, Theora/Speex
MP4 -> HE-AAC/AAC H.264
MKA -> Any audio codec
MKV -> Divx / Xvid/ Seperate Audio Codec
AVI -> Should just dissappear.

A prologic decoder doesn't use any ancillary data. That is a big difference.

Not when u´re getting metallic artifacts from the one that DOES use auxilliary data.

We want to utilize in the future other audio coding standards (e.g. Aud-X method of coding spatial information can be married with Ogg Vorbis and Mpeg4). These will be future implementation steps, if our codec gains interest.
---
Dolby prologic originally is an analog technology of encoding signal and has several limitations (e.g. you cannot encode all 5.1 channels), some spectrum distortins are present, etc.
---
However we want to include dolby prologic decoder to our DirectShow filter,
to enhance its flexibility.
---
We considered several methods of dealing with cross talk, that is why it is much more effective than in mp3surround
Beside much better sound quality, OUR CODEC IS FREE for everyone, and flexible software tools are also available...for free (such as the complete frontend, ACM encoder for VirtualDubMod and DirectShow filter for real time decoding).
In fact, the decoded signal can be (on the fly) encoded to AC3 and sent through SPDiF to an external home theatre equipment... And I think this is very useful feature
---
The idea was to support as many containers as possible.
So Aud-X can be used in Ogg, Matroska, etc....but also AVI
---
I honestly do not know what metallic reverb you are talking about...
Plese let me know what sound track did you test, and I will focus on this.
During our tests everything was fine (beside the expected crosstalk ofcourse).
--
Please compare our 128 kbps STDQ for mp3surround 192 kbps and you'll see the difference.
--
It would be great to test our technology in an official subjective test, with reference
to competitive solutions
(are there any organizations interested in such?)

I'd still like a statement about decoding specs availability, please.

I am sorry, I forgot.
What do you mean by "decoding specs?" avaiability.
Please explain.

Ability to implement a decoder, not just using your one.
It could be either a syntax and decoding description (like ISO mpeg specs), or a sample decoder source code.

Many codecs are successfull because of interoperability and open specs.

Yes, we are planning to provide a source code or a library in the future
(probably within next few months). So the complete SDK with
examples how to use it should be available soon.

But you will need to provide decoding source code, not just source code of an app calling your decoding functions.
Without that, you are unlikely to succeed.

MPEG does exactly that, but you're a little late for the MPEG Spatial Audio standardization.

A test against MP3Surround could be done here now, or a test against MPEG SA later. Even testing against HE-AAC would be interesting. But for 5.1 tests there will probably a bit fewer participants - most people don't have 5.1 systems.

Yes. Ofcourse. I mean source code or at least DLL.
By SDK I mean a VisualStudio project showing how to decode Aud-X stream
using our source (or library).
----
Well, so far 5.1 is not so popular, but I think it would be very interesting
to test it against mp3surround.
And against HE-AAC as well.
----
How could I help in this?

I hate to jump in the middle of the discussion, but you should be aware, as Garf said, that MPEG already is deep in phase of standardising spatial-audio coding (SAC)

RM0 listening tests: http://www.chiariglione.org/mpeg/working_d...ening-tests.zip

The work is already very progressed, so I am not sure if any new technical proposals might get in (although it is still in RM phase), but you can surely try.

Thanx for this information. Interesting document.
Especially that the only sounds used were music.
--
I have only tested and tuned our codec on movie sound tracks mainly
because of one very important reason
(I started to play with multichannel music, but I decided to focus on movies only).

Very often in movie sound tracks channels contain the same sound components, but of shifted phase.
For example in movies like "Fifth Element" nice ambient spatial feeling at the begining was achieved based on phase shifts of some spectral components.
PHASE REALLY MATTERS FOR THIS MASTERPIECE OF 5.1 SOUND,
and as well for other movies with good surround audio.
-----
This makes coding spatial information EXTREMLY difficult, because
merging channels ends up in loosing the phase relations. So after reconstruction
even when you reproduce energy relations between channels, phase gets completely distorted (I mean the phase shift of spectral components between channels is lost).
I tried to show it in my whitepaper when I did the objective test
(I know this is not an appropriate way to compare codecs and subjective tests are much better, but this was the only thing that we could produce at that time).
And it shows phase and energy erros.
Mp3surround is really poor for preserving phase.
So when you watch movies like "Fifth Element" on high quality
5.1 home theatre system, really alot of the spatial feeling is gone...
And we did everything to avoid this!
---
So how is it with multichannel music?
Why in this document are no movie sound samples?
Maybe AAC is dedicated mostly to music, similarly to mp3surround?
Please explain.

The decoder has to be source code. Forget about DLL. It is not just Windows people use, you know.

I forgot to summarize
Our codec is dedicated to movies.
That is why we wanted it to be compatible with AVI,
and that is why we provided DSfilter used mostly by
avi players.
--
Thus it would be very interesting (for us at least ) to see subjective tests
results for movie sound tracks only!
--
Ofcourse tests for multichannel music would also be interesting,
but as a separate test.
--
Please share your thoughts.

No, it is not - AAC itself is a general-audio perceptual codec, and it should be able to cope with any kind of natural signal, providing that you do it wthi reasonable bit rate.

Concerning phase preservation, there are lots of tools in the current RM0 specs of the Spatial Audio Coding (SAC) standard that deal with ITD/IPD/ICC issues, and it also support so-called "Artistic Downmix", where the downmix is not a mathematical derrivate of the original sound but a producer-generated one, so there is plenty of knowledge invested into up-mapping that to original 5.1 with all the difference there is.

It would be great if you could compare RM0 SAC codec with your solution, to see how they stand against each other - but I am unfortunately not able to provide you with the codec, you would probably have to contact someone of the MPEG members participating in these tests for the demo encoder.



Pretty much anything could be forced-into AVI, even with the expense of killing the features of your codec because of that more-than-decade old, cripple historic artefact of the media container (AVI is not a format)

OK, thank you for the info.
Something called "artistic"
is in fact the most common way
to produce surround sound in modern movies
---
It would be nice if someone objective,
not me or mp3surround, or AAC scientists,
would perform such subjective tests.
And IMO separating movie sound track case
from 5.1 music would be of importance,
otherwise the results could be a bit cloudy.
---
Well, AVI is the most popular...crap

I´m uploading two samples, one AudX STDQ and one HE-AAC Tape.
It´s a "metallic aftertaste" I get on the AudX sample, u can hear it in the cymbals and the voice.
5.1_AudX_STDQ.mp3
5.1_HE-AAC_Tape.mp4

BTW, it took over the decoding of AC3 for wich I use AC3Filter. I did not apreciated that since I love AC3Filter and wouldn´t change it for a decoder that lacks of so much options for decoding. I had to unregister the audxac3directshow thingy since merit change didn´t work.

Hi, thank you for the test.
I am working on the DSfilter.
It is improved now, because it used to replace FFDshow.
Now the new installer does not do that.
The next step will be compatibility with AC3filter.
Thanks for this suggestion.
---
BTW. Can you submit the original sound fragment?
The link to Aud-X sound is not working...

As in the original WAV? cuase it´s a 66 Mb WAV, and I have a freaking 56K modem, it would kill me to upload something like that.


R U sure? I checked it was working fine before posting and had no problem. The direct link is "http://mcaudio.byethost33.com/samples/5.1_AudX_STDQ.mp3". It´ll send u to a page where it says "View Image/Download File". That should allow to download it.
Copy/pasting the address works fine too.

I don´t want to have to unregister the DS filter but AC3Filter is by far my choice for AC3 and DTS, and the "taking over" not manageable with filter merit it kind of a downside.

Yes you are 100% right. The Aud-X DSfilter should not replace
the AC3filter. We will provide a new install with this problem fixed.
-----
Regarding the test:
I have an idea.
Please provide an AC3 640 kbps (for example 10 seconds)
sample.
----
So the input will be the AC3 file (reference sound).
And I will be able to download it and will decode it on my computer.
Thank you.

OK, I downloaded both files succesfully.
The original file must be have been AC3, because
this is from a movie. Can you provide a link to the original
(so I will be test all the examples on my home theatre system)
---
How can I convert the mp4 file to 6 channel wave?
Besweet generates 10 kb file as the output...
Please help.

You could try using foobar2000 for that.

(btw I hope you didn't do listening tests against ac3 encoded by besweet?)

Use aacenc32.exe with nero´s dlls, with the -x option it decodes the stream. Or FAAD.
As soon as i can i´ll upload a take from the original .ac3 file, since I find it not practical to decode to wav and reencode to .ac3 again from 448 to 640.

Hmm... I do not have Nero...
Is there any FREE decoder available?
------------------------------
Anyway, yes you are 100% right,
the original is the best choice (AC3 448 kbps)
if available.

You do not need Nero for decoding, there are dozends of free MP4 decoders like foobar2000 or FAAD2.
If you want to encode, you can use the free Nero trial.

OK, thank you.
In fact I want to convert the mp4 file to 6 channel PCM.
Is it possible with Foobar2000?

Got to http://www.rarewares.org/aac.html and get the latest faad2, and get the Ivan&Menno GUI (just to simplify things, not mandatory) from http://members.home.nl/w.speek/ivan&menno.htm.

The AC3 source is up now. I just cutted a 30 secs take from second 90 to 120 (where the artifacts are most noticeable). It corresponds to the last 30 seconds of the samples.
http://mcaudio.byethost33.com/samples/5.1_AC3_448.ac3

OK, I succeeded in reproducing the sound on my sound card.
I decided not to generate the 6 channel waves.
---
My observations:
1) Yes. For this short fragment there are more significant artifacts for Aud-X than for HE-AAC.
2) For some other fragments Aud-X performs better (center channel for HE-AAC has
unpleasent noisy artifacts.
----
3) In general, all differences are difficult to spot (i.e. I had to listen to pairs of channels with my AKG K-401 headphones). But testintg surround sound on headphones is kinda silly
---
When listening to both on my home theatre system, honestly I could not hear any difference.
==================================================
But the most important issue is that this is a very poor example of 5.1 sound!
Sounds like originally it was 3 speakers (with center separated)...
In surround channels nothing interesting is going on (they are very much the same
as the fronts).
Subwoofer is not a special effect, but low pass filtered fron channels...
ETC.
----------------------
I think that it would be interesting to perform a comparison test (subjective plus
some phase and energy distortion measurements) on some high quality 5.1 sound
- originally mastered as 5.1 (surround).
- Maybe someone will decide to perform such a test.
------------------------------------------------------------------------------------------
Thank you for your example anyway.

Here is a nice 5.1 surround sound sample (PCM)
(from Fifth Element)
www.medicalgorithmics.com/zzz_fifthelem_orig48.rar
--
Could you please downoad it and encoded with HE-AAC 128 kbps?

After downloading, please let me know and I will remove it from
the server

Thank you.

So basically you`re saying that your codec only performs well on "cool" 5.1 surrounds, but for normal 5.1 surround that u can find on every movie in the universe, it doesn`t? That is a very lame argument.
If it`s gonna be good, it better be good always, not only on "super" surround tracks.

Please, do not get irritated.
I am saying that the difference is difficult to spot based on this example.
---
Maybe it will be easier with the use of more interesting 5.1 sample
(maybe HE-AAC will show its superiority... I do not konw).
Therefore I provided an alternative sound,
because I am curious of the results.

Don´t worry, i´m not irritated nor I have any bad disposition to your codec capabilities, I´m in no way trying to trash-talk about it, just that the arguments u posted didn´t do it for me.
I´m all forth new codecs or new improved implementations of the existing ones, otherwise I woulnd´t have taken the time to test it.
I just think that is very soon to state that your codec is superior to rest without having done the proper testings within the audiophile community.
The testing file was chosen at random, was the one I had in front of me at the time I read about the codec for the first time.
I still think that for a 128kbps 5.1 file it sounds really good (considering that the source file is in WAV size 67 Mbs big and compressed to 1.8 Mbs).

OK. Thanks for explanation.

Peace, 3d.

We have just lounched Aud-X forum
www.aud-x.com/forum

So everyone who has qustions, suggestions,
requests, bug reports, etc. related to the project

is welcome to share her/his opinion

Short info:
Aud-X MP3 5.1 support added to FFDshow
http://bittekeinspam.googlepages.com/
Thanx

Big kudos to Jacquers
for (unofficial) AVI.NET
support for Aud-X MP3 5.1
http://forum.doom9.org/showthread.php?p=826587#post826587

Just a note for those making the test files:
The levels are not consistent from the original AC3 clip to the other formats. For any real kind of comparison to be made, the unprocessed output of the source should be used [for encoding]. The current tests are more "let's see how bad the generation loss is after post-processing and transcoding of an AC3 stream" than "let's compare encoding formats".

A better sample would be something taken from a PCM source, not AC3.

I have such encoded tracks if anyone is interested in doing a comparison between AAC, AC3, DTS, &c. based on source material that does -not- come from a pre-encoded format. These are from a 5.1 "radio drama" I engineered a couple of years ago that features some very "busy" sections with both sound effects and music.

Last note: while I understand the perceived need for a low-bitrate surround audio format, I honestly have little faith [even after having heard the samples, which work reasonably well] in any format that does not use discrete channel encoding. I already use both AC3 and DTS audio in AVI containers and have no problem with either [transcoding == bad]. Then again, I also prefer lossless-packing formats for stereo [2-channel] audio.