GUIDE: Stereo to 5-Channel Surround [Archive] - Page 6

kempfand

24th December 2003, 01:18

@daphy: Thanks for the initiative.

As a 1st contribution, I'm adding a newly designed bidule which uses the Panorama-VST. It's currently in the "/Ambisonic/upload"-directory, but you might want to sort it differently.

Some explanations: I think it basically achieves the same as
the 'old' one. Maybe it is slightly better, as it
processes internally as 2nd-order-B-format (within
Panorama), whereas B-Pan "only" does
1st-order-B-format processing. I don't yet have
enough samples to really call this a "firm" statement,
so it's just a working assumption currently.

Some notes to the settings (see 2 attached JPG's for
illustration):

(1) Elevation: I put that from 0 to 0.55 (L & R), to
tilt the soundfield slightly upwards by 33 degrees.
This is because my speakers are all very close to the
floor.

(2) Distance: I out that from 3.70 to 5.05 (L & R),
which slighy enlarges the size of the sound-sphere. If
enlarged too much, the VST calculates too much
reverberation for my taste.

(3) Azimuth: For "L", set to +2.20, for "R" to -0.94.
This only applies to when decoded to Pentagon.

For Square-decoding, it would be +1.57 for "L", and
-1.57 for "R".

Why the change (Square --> Pentagon) ?
- "L": +1.57 --> +2.20
- "R": -1.57 --> -0.94

Remember that we turned the soundfield
counter-clock-wise by 36 degrees, inorder to have it
correctly aliged to the 5.0 (or 5.1) speakers in people's
living-rooms.

We have to do the same here, i.e. rotate
counter-clockwise by 36 degrees, which corresponds to
an an increase of +0.63. Meaning:
- "L": +1.57 + 0.63 = +2.20
- "R": -1.57 + 0.63 = -0.94
Comments & suggestions welcome here in the forum.

Happy Xmas to all
Andreas

P.S. joshbm: don't forget us ... ;)

Umma

24th December 2003, 01:45

Some notes to the settings (see 2 attached JPG's for

Um...what two attached jpgs? :)

kempfand

24th December 2003, 01:51

The jpg's are within the _Panorama_-_Pentagon_.RAR I uploaded.

Andreas

Shayne

24th December 2003, 15:30

Just a follow up since i rambled on about this problem for so long in this thread. :o

It appears that the clicking problem has been fixed in Besweet by DSPguru in the latest b24. Clipping at the first .002 seconds on 32 bit float waves is gone.

Since there is two versions of the latest beta you need the latest.

http://forum.doom9.org/showthread.php?s=&postid=417052#post417052

or home

Edit:

With regard to the process and the guide. I find the LFE input nice on dts and therefore have been using it. I remember the warnings about using it in the beginning on this thread.

What i have found is you need to watch to make sure that the lfe channel is not the controlling wave in the normalization process. This is the case with the graph suggested. This is dangerous and really does not normalizes. What i have done is add gains to the lfe channel between HNM and the recorder (3 would be required). I have found the default -3.33 db is a good setting but my testing is young (ed : -5 in most conditions?). This ensure that the normalization is carried out on C, F, or rears and not "pump up the volume".

Peace

kempfand

26th December 2003, 21:24

From Shayne Dec 11: What we need is all this put together in a MP3 player that joins all these filters and outputs a play list to a temp.ac3 file in small chucks and then sends them out spdif. The 3 gain filters can be user adjusted and voilà a player that would definitely out play anything we have now. I am sure its thought of and in the works.
I think we are basically there. :) At least for "wav/mp2-in" --> "6-channel-out", it works beautifully here with Bidule and SIR (Super Impulse reverb) (http://www.knufinke.de/sir/index_en.html). :D

A few (selected) details on the SIR-VST: - This is a reverb device that works with "impulse-responses".
Impulse-responses are generated by recording short, broadband
signals and their corresponding room-reflections, and by cutting
out the initial signal. What is left contains the information
needed to reproduce the same room with other signals.
- SIR is a stereo plugin. To take full advantage of
stereo impulses for mono signals, you must first convert
the track to stereo.
- Features:
- performance optimised
- wave-file import 24-32 bit
Angelo Farino already published an outline Conversion between UHJ and B-format using SIR (http://www.ramsete.com/aurora/conversion_between_uhj_and_b.htm#Audiomulch) for AudioMulch, which is pretty self-explanatory, so I wont' dulpicate here.

I've written a small bidule which does the same as outlined Audiomulch contraption (I can upload it to Daphy's ftp if there is interest).

In summary, this achieves the same as the initial Ambisonic Guide by EoH (http://forum.doom9.org/showthread.php?threadid=29277) , but with the new tools in real-time (or "hyper-speed" with the offline-mode).

A few notes:
- make sure you use the "right" impulses with your source (i.e. 44.1kHz 32-bit pulses on source=stereo_wav with same properties)
- "M2Buss" in AudioMulch translates into "AudioMatrix" for Bidule
- I personally prefer Bidule, but this is my personal taste (think it's more intuitive to use, more straight-forward)
- Filtering-impulses are also available from A. Farina's Site (http://pcfarina.eng.unipr.it/public/B-format/UHJ-conversion)
- You might want to play with the settings for 'dry' and 'wet'. As a starter, I recommend 'dry'= 0 dB, and 'wet0= -6 dB.

Cheers,
Andreas

puzio

26th December 2003, 21:55

Hello Kempfand.

Thank You for your work with Panorama Vst plugin. Can You upload Bidule with SIR VST Plugin?

I think "initial Ambisonic Guide by EoH" gives the best DD5.1 output,
How do You think?

THX

kempfand

26th December 2003, 22:25

Ratings:
- I like Panorama slightly better than B-Pan & B-Proc
- Aurora gave excellent results, but I only used it on 3 CD-conversions, which for me is not enough to make a statement "it's better" than B-Pan.
- However: Given the speed-advantage of the Bidule-setup, I'd clearly give that more 'ranking-weight' than the "maybe slightly better" of Aurora.
- SIR (Super Impulse Reverb) -VST: I've just started to play with that and need to do more tests before I can make a statement.

Andreas

puzio

28th December 2003, 19:19

Hello Kempfand.

Thank You for your work.

Do You see that in Farina's guide with Audiomulch Y-signal is connected with 4-input of B_Decoder , NOT 3-input of B_Decoder.
So We have W,X,Z(silence),Y !!!!! in input of B_Decoder, NOT W,X,Y,Z.

In Bidule as You wrote I did convertion with "Bette Midler - Wind Beneath My Wings" I get "Front Left" loud, but "Front Right" is very low. With "Panorama" way they are similar.

?????

THX

kempfand

28th December 2003, 20:04

Hi puzio,
Do You see that in Farina's guide with Audiomulch Y-signal is connected with 4-input of B_Decoder , NOT 3-input of B_Decoder.
So We have W,X,Z(silence),Y !!!!! in input of B_Decoder, NOT W,X,Y,Z.
Good catch. I personally believe it is a mistake on A. Farina's page.

As you correctly note, the reverb is done to create W,X,Y. No need for Z (as there is no information on Z in the starting stereo-wav).

On the B_decoder, the pin-sequence is W, X, Y, Z (as you correctly note).

According to my understanding, reverbed W should connect to W-pin of B-decoder, X to X, and Y to Y. This is also how EoH's original guide works.

I get "Front Left" loud, but "Front Right" is very low. With "Panorama" way they are similar. I get perfect reults here :) (both with SIR & Panorama). I believe that this problem (and similar ones, such as 'wrongly' rotated sound-fields) are related to how the channel-mapping & demuxing is done.

Some suggestions:
a) Make sure B-Proc is configured as in the guide (http://home.wanadoo.nl/appyhappy/19.jpg)
b) Use BeSweet to demux the 6-ch-wav into the mono-wav's. If you use wav2wav6.exe, you need to either re-name the channels, or use a different channel-mapping.

Good luck,
Andreas

Umma

28th December 2003, 20:30

Has anyone every contacted Angelo Farina about us using his guidelines over here?

kempfand

28th December 2003, 20:42

I'm not sure if I read your question ('his' guidelines ?) correctly (bear in mind that my native language is not English).

- If you refer the puzio's question about the wireing of the reverbed W, X & Y, then the answer is no (it's on my todo list).

- If you refer about potential 'ownership' of guidelines, then let me state that Farina's page was just recently updated, and that the bidule I outline above does exactly what what already described here in July 2002 ( A Guide from Stereo to 5 channel Surround ! (http://forum.doom9.org/showthread.php?threadid=29277) .

As I said: My native language is not English, so I might mis-read.

Andreas

puzio

28th December 2003, 20:59

Umma

28th December 2003, 21:35

Originally posted by kempfand
I'm not sure if I read your question ('his' guidelines ?) correctly (bear in mind that my native language is not English).

- If you refer the puzio's question about the wireing of the reverbed W, X & Y, then the answer is no (it's on my todo list).

Sorry about the ambiguity. I should have been clearer as the wiring would have been a topic to touch with him, that and...

- If you refer about potential 'ownership' of guidelines, then let me state that Farina's page was just recently updated, and that the bidule I outline above does exactly what what already described here in July 2002 ( A Guide from Stereo to 5 channel Surround ! (http://forum.doom9.org/showthread.php?threadid=29277) .

I know it was with a couple of others (and you, too, right?) that EoH put together the first stereo to 5.1 guide, but he also referenced Farina's site in the use of Aurora to create the b-wave. I'm not saying Farina 'owns' the method of conversion, but by providing us with directions in using Farina's Aurora program (and with a big nod to Farina's own guide to using Cool Edit and Aurora), EoH indicated that Farina was a big inspiration at least in the beginnins of the conversion process. Of course it is a dynamic thing and it will constantly be changing as newer stuff comes out, but I guess my question should have been: Has Farina ever seen how his suggestions and use of his Aurora software have inspired people? Or something like that...:)

I didn't mean any offense, Andreas. I was basically wondering if anyone had contacted him about anything over the last year or so.

Pax,

D

kempfand

28th December 2003, 22:20

Has Farina ever seen how his suggestions and use of his Aurora software have inspired people? My guess is no. I'll write him and give the background. I've been a bit sloppy here and should have done it earlier. I agree that he should be given the right credit.

// off-topic on //
Reason I didn't do it yet: I've sent a really nice & kind email to Dave Malham, saying "thank you" for the nice tools he and his team provide for free (B-Pan, B-Proc, etc). Never heard back a single word ... Needless to say that I was very disappointed (fully understanding that he might have personal reasons which make him focus on other things). Similar experience (on different topics) with Mark Taylor on tooLame (exe/dll).
// off-topic off //

Peace and kind regards,
Andreas

Edit/Add: Also wrote an email to Christian Knufinke, the author of the SIR-VST, saying "thank you". Without the effort from people such as him, all this wouldn't be possible.

Eye of Horus

29th December 2003, 14:22

Originally posted by kempfand
My guess is no. I'll write him and give the background. I've been a bit sloppy here and should have done it earlier. I agree that he should be given the right credit.

// off-topic on //
Reason I didn't do it yet: I've sent a really nice & kind email to Dave Malham, saying "thank you" for the nice tools he and his team provide for free (B-Pan, B-Proc, etc). Never heard back a single word ... Needless to say that I was very disappointed (fully understanding that he might have personal reasons which make him focus on other things). Similar experience (on different topics) with Mark Taylor on tooLame (exe/dll).
// off-topic off //

Peace and kind regards,
Andreas

Edit/Add: Also wrote an email to Christian Knufinke, the author of the SIR-VST, saying "thank you". Without the effort from people such as him, all this wouldn't be possible.

I mailed Farina for at least 3 times and never got a response back !

grtz,

EoH

Eye of Horus

29th December 2003, 14:40

Originally posted by Umma
I didn't mean any offense, Andreas. I was basically wondering if anyone had contacted him about anything over the last year or so.

Pax,

D

I try to contact him long before I wrote my FIRST guide with the Aurora plugins. Because he didn't answer any of my emails, I decided to go on without him and luckily I found someone who helped me in the process.

The second guide was written without any of his software.

This year I emailed him for the third time and again got no response.
I gave up :-)

To me it's not quite clear where the problem is...... Guidelines like Farina's are on so many sides, that it's still the question if these were his' in the first place ! I mean.... he didn't invent Ambisonics ! And everyone can write guidelines about how to use tools. (even I !!)

Don't get me wrong : the guy is a genius, but I don't think it's necessary to "ask him permission" to write guidelines !

I would have loved however if any of these guys came on this board and supplied us with more tips and help. But..... after emailing (Andreas did some emailing too !) and no replies, we gave up hope...

What rests is that we have to find it all out by ourselves :-))

CanI add a nasty one ? If not allowed the moderator can remove it : I am not surprised that when people are selling stuff, but don't reply to emails, the potential buyers will look for illegal versions. If I ask if a routine can be explained, before buying and I get no answer, I won't buy it ! But I will try to find it out by myself with a not so legal version ! Seeing the enormous response my 2 guides have on this board, the makers of these software should think again. That's why I will give 10M kudos to David from Plogue ! When he start selling, I want to be one of the first to buy his program !! And...... don't forget it works two ways. We act as his beta-testers :-)

my 2 eurocents !

EoH

kempfand

29th December 2003, 16:47

Great news: I've just got mail from Angelo Farina :), which I share with his permission. Subject: Re: UHJ t B : VST-method using SIR & Bidule

Hi,

I have seen Your postings, and it was evident that in my original AudioMulch
patch there were severe errors (patching errors and improper gain settings
in SIR). So I corrected everything (I hope)...

I also tested the system with Bidule, and added a corresponding image
on my web page:
Also usage of Audiomulch/Bidule and SIR (Super Impulse reverb) are explained (http://www.angelofarina.it/aurora/conversion_between_uhj_and_b.htm#Audiomulch)

I also saved the Bidule patch on my Public server area: Link (http://www.angelofarina.it/public/B-format/UHJ-conversion/)
Please, check if everything now is OK....

Regarding the concern about the usage of UHJ-to-Surround conversion for
not-UHJ stereo recordings, I can explain what follows:

1) Most coincident or near-coincident (i.e. ORTF) microphonic techniques
still work reasonably well in a UHJ-decoder, as "artificial" mixing (as
most pop/junk music is) designed for Dolby Surround (as almost anything
found on VHS tapes or stereo DVD-video). Spaced or binaural recordings,
instead, are not optimally treated by UHJ decoding networks.

2) The decoding impulse responses DO NOT CONTAIN any room information: the
surround channels are decoded from the original recording only, and contain
information which appear "out of phase" on the Left-Right channels. This is
very similar to the "Dolby Surround" approach, but here the result is stereo
instead of mono, and is full-band instead of band-limited. This outperforms
original Dolby Surround, Dolby ProLogic and similar 2-to-5 channels decoders
(Circle Surround, SRS, etc.). The only serious competitor which can rival or
even sound better than UHJ decoding is actually Dolby Pro Logic II
(available only in very recent surround processors).

3) For very dry original recordings there is no "surround" to be extracted
at all, and none of the above-mentioned methods (nor UHJ) can work. The
better solution, in these cases, is to employ convolution with room reverb.
I suggest, in these cases, to employ the Ambiophonics decoding method
(www.ambiophonics.org (http://www.ambiophonics.org/)). Similar results, but less mathematically rigorous,
can be obtained with some Lexicon hardware tools (such as the 960L).

You can, of course, post this info also on the doom9 forum....
Bye!

Angelo Farina

The mentioned Bidule is correct. Based on my and EoH's listening tests, which compared different decoding-layouts (Square, ITU 5.1 loudspeaker, Pentagon) and different VST-decoders (B-Dec, Emigrator, VVMic), we prefer the Pentagon-layout, but feel free to make your own tests.

Kind regards,

Andreas

Umma

30th December 2003, 16:50

Wow. What a response. At least we know he read these posts. Strange timing on my part to ask about contacting him...

I did some comparisons with 5.1 and Pentagon, and Pentagon wins out, too. Not just with me, but to about three others who listened to some Simon and Garfunkel that I did. Pentagon won.

I really don't know what is meant by "dry" recordings, though. :confused:

And it is still not clear why the ambiophonics method should be used. I read through some of the stuff on the web site, but then I think "ambiophonics," I think about what Jonas was saying about using a room impulse created in the listening environment...I don't remember seeing anything about that at the ambiophonics web site.

specise_8472

30th December 2003, 18:36

Have just uploaded to the FTP site a VST plugin to do ambiophonics.
At the moment just does the six mentioned encodings.

On a lot of recordings there is a big difference from ambiosonics decoding. One I tried it out on was The Music Of The X_Files. Track 6 contains chiming bells. The reverb through the rears was spectacular. Sounded good in Emigrator too, but was not quite as pronounced, and dropped off sooner.

Ambiophonics was produced as a way of producing true 5.1 from the information in an WXYZ file. This file contains the inforamtion needed for decoding
X = Front back
y = Left right
z = Up down (when included)
w = pressure signal (basicaly the input sound at -3db.

Any suggestions improvements etc. Let me know.

On another note, I have just got Roomverb M2 from www.spinaudio.com
Putting it between the input signal and B-Proc produces very good results. This is a very good reverb engine that as heaps of presets and full control over every setting. Some recordings are really brought to life. There is a demo avaliable.

kempfand

30th December 2003, 22:01

specise_8472: Thanks for sharing this. This looks really :cool:

It's funny: Yesterday, I was just reading the corresponding paper (http://www.ambiophonics.org/AESJune2003/AES_June_2003_3.htm) and thought this would be great to try. I don't habe the necessary coding experience though, to put this into dll.

I will try it tomorrow :-)

Thanks again and kind regards,

Andreas

kempfand

31st December 2003, 01:10

Update: Specise's ambiophonics-VST ROCKS ! I just did a testing series with one song (Armik: Rubia; guitar music).

Created 9 tracks as follows:
1) B-Pan, B-Proc, Emigrator (Pentagon)
2) Panorama, Emigrator (Pentagon)
3) SIR, Emigrator (Pentagon)
4) Ambio I
5) Ambio I~
6) Ambio J
7) Ambio J~
8) Ambio K
9) Ambio K~
All were batch-BeSweet'ed to dts with normalization (@DSPGuru: BeSweet working in batch-mode like a charm).

Based on this one-song-only-test, B-Pan is very good, but for my ears, "Ambio J" wins (more discrete impression, nicer & broader sound-image). In general, I preferred the I , J, K -type over the "~"-type, but this is my personal impression and based on 1 initial song only.

In summary @ All: Highly recommend you give it a try.

If you want more technical background, read the corresponding papers from Robin Miller (http://www.filmaker.com/) , especially
- Transforming Ambiophonic + Ambisonic 3D Surround Sound to & from ITU 5.1/6.1 (http://www.filmaker.com/papers/AES114_Transform.pdf)
- Scalable Tri-play Recording for Stereo, ITU 5.1/6.1 2D, and Periphonic 3D (with Height) Compatible Surround Sound Reproduction (http://www.filmaker.com/papers/AES115_Scalable%20Tri-play%20Recording.pdf)

@specise_8472:

Any suggestions improvements etc. Let me know. I'm just thinking on howto achieve a workable & affordable solution to include the SC (surround Center) speaker. My amp (Yamaha 630) has a speaker-connection for SC.

If I understand this correctly, SC with dts is matrixed into the SL&SR. But this is currently probably a 'dead-end' as I'd guess that no affordable encoders are around.

Any chance or idea on howto 'achieve' this packaging of SC (into SL&SR) already within the VST ? Just thinking lound ...

Kind regards,

Andreas

specise_8472

31st December 2003, 04:46

Originally posted by kempfand
Update: Specise's ambiophonics-VST ROCKS ! I just did a testing series with one song (Armik: Rubia; guitar music).

I'm just thinking on howto achieve a workable & affordable solution to include the SC (surround Center) speaker. My amp (Yamaha 630) has a speaker-connection for SC.

If I understand this correctly, SC with dts is matrixed into the SL&SR. But this is currently probably a 'dead-end' as I'd guess that no affordable encoders are around.

Any chance or idea on howto 'achieve' this packaging of SC (into SL&SR) already within the VST ? Just thinking lound ...

Kind regards,

Andreas

As is mentioned in the papers, 6.1 AC3 is matrixed center surround, and so does not work properly. At the moment I have done as suggested and mixed the SC channel into the LS RS at -3db. The only way to do a proper SC is to use DTS Discrete 6.1 which there is currenly no software for that I know of. Nuendo and Surcode only support 5.1.

I can output the SC if wanted, if you have a reason for it?
I am looking at writing a PLI/II encoder in the future to use. Also I am looking now into a filewriter that thakes the output from the Steinberg VST AC3 encoder and puts it to disk as a valid AC3 stream. Just think from Source to AC3 in one hit.

Also in the papers are mentioned that you can recover the original WXYZ file from the 5.1 track. Would there be a use for this facility to turn original AC3 tracks (and DTS) into WXYZ files for processing through emigrator to more or less speakers? Just need to reverse the plugin and change the co-efficients.

About the difference between ijk ijk~, the ~ are basically the same but with the microphone tilted -30degrees.

Just thinking out loud:D

Ace_V

31st December 2003, 22:16

Hi,

I think I've set up every thing correctly but my front and rear channels sound very similar. There shouldn't be as much vocals in the rear channels as in the front, right?

Can anyone help me? I must have done something wrong.

Also just tried it with a mono file which I would have expected to only come through the center channel or at the most the front channels, but it has pretty much the same sound in all channels except the LFE.

thanks

kempfand

2nd January 2004, 02:36

The only way to do a proper SC is to use DTS Discrete 6.1 which there is currenly no software for that I know of. I had to do some reading an am still getting 'up to speed' on the topic. Regarding dts ES, I understand there are 2 flavours: DTS-ES Discrete , for which I also understand there is currently no encoder, and DTS ES matrix. See e.g. technical_literature (http://www.dtsonline.com/media/uploads/pdfs/DTS-ES%20Overview.pdf).

For the "DTS ES matrix", I understand all that is needed is:
(a) 45 degree phase shift applied to LS and RS signals
(b) CS at -3db
See above source, as well as dts-es.pdf (http://www.tronika.no/support/files/dts-es.pdf)

Update / Edit added below

Now: As is mentioned in the papers, 6.1 AC3 is matrixed center surround, and so does not work properly I didn't find that concrete recommendation :mad: (can you provide a link, maybe I was just having bricks in front of my eyes). If it that's what is stated, we should test. I'm stating this in analogy to the 5.0-decoding rigs (Pentagon vs. ITU 5.1 layout), where 'recommendations go for 5.1, and personal tests often go for Pentagon.

---
Would there be a use for this facility to turn original AC3 tracks (and DTS) into WXYZ files for processing through emigrator to more or less speakers? Still thinking about this. I could see a potential application for 'down-mixing' into 2.0. Would be keen to see how this compares to other methods (e.g. azid).

Don't want to 'make you work', but I'd be very keen and happ to test, if it's done without too much effort (realising here that nothing is 'easy').

---

Regarding PanAmbio 4.1: Robin Miller (http://www.filmaker.com/2dsurround.htm) mentions this as For a single listener, PanAmbio is superior to 5.1 in accurate 360° localization, spatial impression, and envelopment - a benchmark of excellence Any idea on howto achieve this witha bidule ?

---

All this is also thinking out loud :D

Kind regards,

Andreas

P.S. Have done more tests with your Ambi-rc1.dll. It still 'rocks' and wins !

Edit: I'm assuming that "DTS ES matrix" can be encoded with current SurCode version, if conditions (a) & (b) above are fullfilled.

I did a test as follows; BeSweet/azidts-demux a "DTS ES matrix" encoded audio (had to use "-ota( -fs 44100"), and SurCode-re-encoded. FOrcing "DTS ES Matrix" on the amp resuited in the same audible siganl in each speaker (also SC)

specise_8472

2nd January 2004, 03:59

Okay, one at a time

For the "DTS ES matrix", I understand all that is needed is:
(a) 45 degree phase shift applied to LS and RS signals
(b) CS at -3db
True, but Surcode does not allow a true implementation of this, as in it will encode okay, but will not set the flags needed in the frame headers to indicate that the front or/and back left-right pair/s are matrixed encoded with the appropriate center signal. Writing a program to go through and do this after could be done. But I have found that the published specs and the actual specs are just different enouth to cause trouble on this point. There are unknown flags that are being used:(

For a single listener, PanAmbio is superior to 5.1 in accurate 360° localization, spatial impression, and envelopment - a benchmark of excellence
I have not found anything majorly useful at the moment to look a adding this feature (yet:D). I am still testing/playing around. A lot of this centers around crosstalk cancellation, which I am investigating.

P.S. Have done more tests with your Ambi-rc1.dll. It still 'rocks' and wins !
As Robin Miller points out, Ambisonics suffers from lack of spaciousness. But a hybrid system of the two prtoduces the best results. Basically what I am trying to acheive. Per-Ambio from Ambisonic WXYZ. Also Ambiophonics really only caters for the single listener sitting in the "sweet" spot. Whereas Ambisonics has a large "sweet" spot.

specise_8472

3rd January 2004, 02:22

Have just uploaded Version 1 of ambiophonics decoder.
Have added 6 new presets, and deleated the 3 that did not work good.

The new presets are based upon proper Ambisonic decode equations, the two versions of each are s = spherical decode and p = controled opposites.

Attention Daphy - just noticed that my old DLL was gone from the site. Was this you, or are you going to have the age old problem of people deleating stuff for the hell of it:angry:

@ndy

3rd January 2004, 02:56

@specise_8472
Attention Daphy - just noticed that my old DLL was gone from the site. Was this you, or are you going to have the age old problem of people deleating stuff for the hell of it

That was Daphy and me :D

So what I can say is that we have to make some rules for the FTP- Sever.

If your file was named "Ambi-rc1.dll" you`ll find it in "\Ambisonic\VST-Plugin\plugins-031231.rar".

The best way to get along is, that all of you upload the files with your nicknames. Instaead of zipping the files there should be a categorized folder in which you can find the uploads as well as an zipped file with all available files in there.

I don`t want you to make double postings, so the threads "stereo to 5 channel surround" and "dedicated ambisonic server online" should be separated.

Feel free to post your uploads of bidules, dlls and somewhat else in the ambisonic post.
All technical problems should be posted in the server posting.

The structure of the FTP- Server is quite easy:
you`ll find all your uploads as soon as possible in the folders named to the subject. I agree that the uploads should not be zipped to find them a little easier (Note: I unzipped the file plugins-031231.rar)

If you have other suggestions of organizing the FTP- Server please let me know :cool:

BTW: except of daphy and me has no one the (official) right to delete anything :D

Last note for this posting:

please upload your files with your nickname in the file file like ²Ambi-rc1_specise_8472.dll".

CU
Andy

specise_8472

3rd January 2004, 12:47

thanks to @ndy I have now uploaded version 1.0 of the Ambiophonic DLL to the server. In a directory named (strangely enouth) ambiophonics.

Take note that the server will be down from 5 - 9 Jan.

@kempfand
hope to upload tomorrow (time permitting, but gotta do things with the partner once in a while:D) the reverse process I was talking about. The easiest way to use it will be to use Hypercubes HYPERCUBE decoder, can decode AC3 and DTS directly to 6 channel wave:)

I will be having the inputs to match its output.:D

kempfand

3rd January 2004, 23:29

@ specise_8472 :

Thanks for the Ambi 1.0.dll :). I will be happy to test the reverse process. Don't kill yourself (know that partner & friends need time too ;) ). A few questions/comments:

(1) With reference to First and Second Order Ambisonic Decoding Equations (http://www.muse.demon.co.uk/ref/speakers.html) I understand what 'Controlled Opposite" and "Spherical Harmonics" means. I'm a bit unclear however as to what you mean with the "O", "P", and "S" -options. :confused: Would appreciate if you clarify a bit.

(2) The easiest way to use it will be to use Hypercubes HYPERCUBE decoder, can decode AC3 and DTS directly to 6 channel wave Preferring BeSweet here, as it well supported and documented :D

(3) Regarding DTS ES Matrix and Surcode does not allow a true implementation of this Agree. That's also why I had to force 44.1kHz in order to BeSweet-demux a '44.1_kHz_DTS_ES_Matrix' (using "-ota( -fs 44100" ). Also saw your previous threat (http://forum.doom9.org/showthread.php?threadid=61203) on the topic. Regarding a proggie to set the correct flags, there was a previous threat with regard to AC3: Downmix 6.1 EX to normal 5.1 (http://forum.doom9.org/showthread.php?threadid=39044), where member 'S_O' talked about a small proggie 'ac3ex'. Might be useful here.

Kind regards,

Andreas

specise_8472

4th January 2004, 01:38

O P S are just Octagon Centered, Pantagon and Surround decode matrix.
The only difference to Emigrator is that Pentagon is different, in that there is no C channel as such. I re-computed for center and also the other channels so as not to need to do a 30 degree rotate to bring speakers in line.
I noticed that the LR channels in these decodings are far to loud and overpower the surround/c channels. Have fixed this and am implimenting a decent crosstalk cancellation on the LR pair.(hopfully). The IJK are unchanged.

About Surcode. As I said then, DIE ANOTHER DAY uses Matrixed SC channel.(Must do as it does not use the extended channel specs - but all other 6.1DTS I have do). But the flags that are in the specs are obviously not the ones used. Have not got around to investigating this properly yet. At the moment when I get time, I will check out the Sum/Difference flags to see if and how they work.

After reading up on DTS ES again, combined with all the reading I have been doing recently on Dolby surround, I have found a new path to follow regarding this. Thanks for pointing me back down this path.

Will follow up next week.

Umma

4th January 2004, 03:18

I noticed that the LR channels in these decodings are far to loud and overpower the surround/c channels. Have fixed this and am implimenting a decent crosstalk cancellation on the LR pair.(hopfully). The IJK are unchanged.

I did a "K" encode of live concert and what bothered me was the rear surrounds seemed too loud. I split the Bidule wav with Besweet into 6 mono files and opened up the files in CEP. The SL wav was HUGE, going to the top and bottom of the scale, and the SR wav was much narrower. I wondered about this, but didn't want to say anything (I am not proficient in audio geekspeak :) ). I thought maybe there was a reason for it (though I couldn't see it).

Kempfan is right: It rocks, dude. Thank you very much! I like the K encode. We really didn't hear a difference between the others you added.

And I really appreciate it, man! :)

specise_8472

4th January 2004, 10:12

You are right. I introduced a big bug when adding the Ambisonics decodes into the mix. For now please keep using the old RC1, which are correct for what you want - IJK. I will add the SC channel into the Ambisonic decodes as well.

As you may have read,I have found how to do DTS6.1 and will be adding it into the mix next release. Actually, looking at my old sourcecode to RC1 I have actually done this mix to create 6.1.

It should work. If not, try putting the streams directly into surcode without any post-processing with besweet. Just split them out unchanged. I have a feeling that things like Besweets Normalization etc may change the signal enouth to confuse the decoder?

As I don't use Besweet, I would like someone to check this out. I have succeded in just resampling from 44.1 to 48.0 to make DVD spec audio okay. But the huge mind-blowing surprise was finding that you can get 6.1 on 44.1 cd audio:eek:
(Now I suppose everyone who wants will have to redo all their hard work:D - assuming anyone wants 6.1 audio cd's;)

Btw I am using a Marantz 7300ose amp for testing. Works perfectly everytime.

@ndy

4th January 2004, 11:56

Originally posted by specise_8472
[B]You are right. I introduced a big bug when adding the Ambisonics decodes into the mix. For now please keep using the old RC1, which are correct for what you want - IJK. I will add the SC channel into the Ambisonic decodes as well.

I searched for the old RC1.
It is now available again in theses folders on our FTP-Server (ftp://daphy.mine.nu) : \ambiophonics and \incomming\ambiophonics.

Have fun

@ndy

kempfand

4th January 2004, 12:22

But the huge mind-blowing surprise was finding that you can get 6.1 on 44.1 cd audio Same experience & surprise here (also played with "007: Die Another Day"). :D

There also is a 6.1 dtswav at SR 5.1 MultiKanal (http://www.sr.se/multikanal/english/e_index.stm) which can be used for testing. On that page, scroll down to "Saga". It says "Coded in DTS-ES, 6.1 Discrete. (All in Swedish)", but my feeling is that it is 'matrixed SC (based on demux and SurCode-reencode).

I have a feeling that things like Besweets Normalization etc may change the signal enouth to confuse the decoder? I always use BeSweet's PreGain Normalization to 100% ( "-ota( -g max )" ), and for me, it worked very good with I, J, & K (RC-1 dll-versions). I realize that DSPGuru recommends using the exact "PreGain Extract Peak chunk" option ( "-ota( -g peak )" ), but I had some problems not doing it "correctly" (which I attributed to the work-in-progress on how Bidule's filewriter actually writes the data).

Regards,
Andreas

Umma

4th January 2004, 15:26

(Now I suppose everyone who wants will have to redo all their hard work - assuming anyone wants 6.1 audio cd's)

SOME of us po' peoples are still stuck in the dark ages and are financially limited to their once-expensive Yamaha 5240 that has ONLY 5.1... :sly:

I'm at work now, and will be for a while, but I have a question. If I add a gain control to the RC1 bidule on the last two going into the sound recorder...the two rear surrounds (so I can lower the rear gains -3dB or so)...AND use Besweet's -ota(-g max), will the resulting mono SR and SL keep their lower gain after splitting the multichannel wav? Or should I stick with -g peak? I only want to do it for some of the live concerts, depending on the clarity of the recording, and this new bidule really adds some nice, startling dimensions to the sound.

Thanks,

D

specise_8472

4th January 2004, 16:39

Have posted V1.5 and bidule to match.
Major change is removed Ambisonic decodes as feel that emigartor does a good job. It is possible for the author of emigratr to add the SC channel into rigs also.
Have added the z input to the mix as I am now testing the ability to add SC to existing 5.1 original mixes by using the reverse plugin (which computes a z channel)and either redoing whole file this way or just using the reversed wxyz to only compute the SC channel and adding it back to the original file.
I have a lot of testing to do yet on different methods:)

On adding gains to outputs - play - and let your ears be the judge. What it sounds like to you is based upon your system (speakers, amp, dvd/cd, cable etc etc etc). Play until you get what sounds good to you.

Attn moderators - is it possible to thread split this disscussion on ambiophonics and my plugins out as I feel it is a different way as opposed to EOH method?

Umma

4th January 2004, 17:36

On adding gains to outputs - play - and let your ears be the judge. What it sounds like to you is based upon your system (speakers, amp, dvd/cd, cable etc etc etc). Play until you get what sounds good to you.

That wasn't what I was asking, for that is what I do. My question was more about Besweet after Bidule, and whether or not Besweet will neutralize any gain changes I make in Bidule. As far as I understand it, both -g max and -g peak both normalize the wavs some kind of way...right? I couldn't find the answer using the search. :confused:

Did you fix the bug in the new release you posted this morning, or should I still use the RC1 release? I ought to be home in about three more hours... :(

Pax,

D

kempfand

5th January 2004, 00:31

@ All: If you want to get the latest Swiss Center / York Ambisonic VST's, go here: Please note - this is a temporary download page (http://www.dmalham.freeserve.co.uk/alpha/vst_ambisonics_alphas.html) . Please reas the info, especially the "You must inform users of its source and acknowledge its use" (which is only fair, as these people really deserve according credit).

@ specise_8472: I like the logo :D . On a more serious basis, does v1.5 decode as 'controlled opposites' (which is what I preper) ? If not, could you add an option ? Not trying to 'ask' here, just want to help to improve, test etc.

Kind regards,

Andreas

specise_8472

5th January 2004, 01:48

On a more serious basis, does v1.5 decode as 'controlled opposites' (which is what I preper) ? If not, could you add an option ? Not trying to 'ask' here, just want to help to improve, test etc.

Controlled opposites only works on Ambisonic decodes. Ambiophonics uses a different method. I removed the ambisonic part out of 1.5. I feel that emigrator does a good job. I might consider doing a different plugin based on Ambisonic alone, with the DTS mixing added in. The SC channel can be added into the set Rigs easy enouth. I am always looking for ways to improve on my plugin. And I removed the ambisonics also so as to concentrate on other things I want in it.

Eye of Horus

5th January 2004, 10:37

Hi all,

I am pretty much behind on the latest developments, due to my illness.
But I have a lot of contact with Kempfand and yesterday we did some tests to compare the Specise Ambiophonics with a new one we found online.
These tests are posted on alt.binaries.sounds.dts.
Download hot.wav from me and the repost hcDTS.wav from Kempfand and compare. Let's see if you all come to the same conclusion as I did....

Now you want to know about this new method, huh ?

It's a VST plugin you can use in Plogue.
The name is CS II VST and you van download a 7 days trial at http://www.srslabs.com.
Don't look at the price :-) Quality comes with a high pricetag !!

Here are the details to make a bidule (the FTP site is in Egypt now..... 0.0 K/s connection :-))

1. Install the plugin
2. Use the decoder
3. Use a stereo file player and a 6 channel recorder
4. The settings of the plugin :
- circle surround : on
- rear center : off (we keep it 5.1 because of the 6 channel recorder !)
- center : on
- post/process : ON (if you want to use True Bass and Dialog clarity)
- True bass LFE : on (5.1) with a setting of 5.0 (this has nothing to do with number of the channels !)
- True bass front : off (other way round for 5.0 !)
- Dialog clarity : on setting : 5.0
5. The pin connections from CS to the recorder :
(first=CS second=recorder)
1-1
2-2
3-3
4 not used
5-5
6-6
7-4
6. Demux with Besweet

To me this sounds as one of the best stereo to surround conversions I ever heard !!
I'm anxious to hear your opinions.........

kind regards,

EoH

specise_8472

5th January 2004, 11:19

I had a look at this a short time ago. As you say good.
I use the Windows Media player version all the time. Does a good job.
If you want, I can make a vst module to mix the outcoming signal into a valid DTS neo6 signal?
I already basically have this as I wrote it to test out my DTS theory.
SC in and mixed with LS RS for output.
Just tell me if you want a particular order of input to output. I can make a simple 3in 2 out if you want.

BTW nice to see you back on deck EOH.

Eye of Horus

5th January 2004, 11:43

Originally posted by specise_8472
I had a look at this a short time ago. As you say good.
I use the Windows Media player version all the time. Does a good job.
If you want, I can make a vst module to mix the outcoming signal into a valid DTS neo6 signal?
I already basically have this as I wrote it to test out my DTS theory.
SC in and mixed with LS RS for output.
Just tell me if you want a particular order of input to output. I can make a simple 3in 2 out if you want.

BTW nice to see you back on deck EOH.

It will take at least another 5.5 months to fully recover, but at least I can concentrate now for more than 30 minutes :-)
There is a huge difference between the WMP or Winamp plug, compared to the VST plug !
BTW I only have and use 5.1, but be my guest and come up with some more good work :D :D

grtz,

EoH

DSP8000

6th January 2004, 02:20

Eye of Horus

6th January 2004, 13:10

Kempfand showed me the channel mapping can be done without my extra step.
This reminds me again that I am not completely healed yet :-)

I edited the message !

EoH

bitsnbytes

7th January 2004, 00:14

Been awhile but hey lifemoves you in different directions heh.

well I just tried the circle surround thing... I'm not saying its bad. but my observation finds that it deminishes the fullness of a song... but yes it does direct the vocals more to the center channel... my test subject was limp biscuit behind blue eyes, so nothing outstanding "musicaly".
I wanna call it a softer take on surround sound.
I did play the reg ambisonic one right after the circle surround... I think that they proly have there best and worst uses each... not considering that everyone has different taste.
I guess it would be the soundstage is larger in reg.... I would not recommend circle surround to any heavy music but for softer music proly a nicer touch. and if you're not doing music with vocals theres no sense in it.

Glad to see you guys are still hammering away at this... my 3 version dvd turned out sweet.. as i did 3 versions of a couple albumns 1 was 2 channel DTS and 1 was 3ch DTS and of course 6 ch dts... took some time not having offline mode but now I see I can get it done quikly ;-)
and almost time to upgrade my rotel:devil: