ffdshow tryout project : HD audio discussion

[SamuriHL]

Thanks SamuriHL for your help offer. So far the closest I've gotten is DTS HD MA with MPC HC 1393 and FFdshow beta 49. True HD is still Multichannel and low audio output. The DTS HD MA is at the proper level. I got this by unchecking allow direct to file output - it was checked by default in FFdshow under output. How is the TrueHD or DTS HSMA chosen? Some disks have lpcm and True or MA. Anyway, great work everyone. What an achievement! Working on an HD PVR now.

I have the allow direct-to-file output checked. The only thing I did was check all the passthrough options. Everything else is default. Welll, ok, not EVERYTHING. I do shut off the mixer. I have 16 bit checked, which is default. I don't mess with the codec options. It just simply works. I don't change the output option, either. I keep everything very simple. It sounds like it's trying to process TrueHD in your case. Are you sure ALL the passthrough options are checked? Including AC3 and DTS.

[davinleeds]

Yes all pass through is checked. All across the top. I have 24 bit checked. I'll uncheck that. I just watched T Salvation while unpacking the HDDVR and at first I wondered if I was getting 5.1 True but then those Skynet thopters showed me the truth. Having direct to file doesn't allow DTS HD. I installed the .1249 MPC and got he same results.

MSI 785 MB 4200 w/5750 W7Pro32

Add not one glitch with anything, nothing. With lpcm. FF would eliminate audio. This worked without flaw. Impressive.

[SamuriHL]

Neat. Well I don't know what that setting does. Nor do I know why it insists on decoding TrueHD for you. I guess post some logs and we'll take a look at them. You're running the 9.11 drivers?

[sub24ox7]

I am going to try a dolby digital plus pass through, I am remuxing national lampoons Christmas vacation hddvd with dolby digital plus to a m2ts. Will post results ,oh and i have a lot of old bd's with lpcm that i can try and log for you if needed.

[htpc66]

It is interesting to note that the new ffdshow beta automatically deactivate "dts pass through" for dts-hdma track and decodes only its core AND no longer crash, horray!

This applies to my Xonar/9800GT and ATI4XXX setups.

Unfortunately for my Xonar/9800GT setup, ASAudioRenderer.dll (original Asus TMT2 version) refuses to load at all even though registered and the relevant registry states "ThreadingModel"="Both".

Any idea truehd/dts-hdma pass through would ever work for other setups like ATI4XXX, Nvidia 8200/8300/9300/9400 and intel graphic chipsets?

[albain]

Sebastii : reviewing LPCM logs, I am not sure we have the same, could you post your sample on mediafire ?

Thanks

Otherwise it seems that the LPCM streams formatted by TMT are 24 bytes long : 24 / 8 channels = 3 bytes = 24 bits

Wherease the LPCM stream formatted by FFDShow are 16 bytes long : 16/8 = 2 bytes = 16 bits

I don't understand (yet) why it wouldn't use the code path where it detects it is a LPCM HD stream and would pad them on 24 bits.

Anyway, as I said the code already exists but it should be fixed for this case and also I had to disable FFDShow resample and mixer filters that will blow off the LPCM.

I think that one option "LPCM passthrough" should be added to avoid FFDShow modifications on those streams (and on the other hand apply the filters on the other streams).

To Sebastii too: about Vista+xonar, we'll need to wait that your AVR is repaired I guess to fix this

[Sebastiii]

Sebastii : reviewing LPCM logs, I am not sure we have the same, could you post your sample on mediafire ?

Thanks

Otherwise it seems that the LPCM streams formatted by TMT are 24 bytes long : 24 / 8 channels = 3 bytes = 24 bits

Wherease the LPCM stream formatted by FFDShow are 16 bytes long : 16/8 = 2 bytes = 16 bits

I don't understand (yet) why it wouldn't use the code path where it detects it is a LPCM HD stream and would pad them on 24 bits.

Anyway, as I said the code already exists but it should be fixed for this case and also I had to disable FFDShow resample and mixer filters that will blow off the LPCM.

I think that one option "LPCM passthrough" should be added to avoid FFDShow modifications on those streams (and on the other hand apply the filters on the other streams).

To Sebastii too: about Vista+xonar, we'll need to wait that your AVR is repaired I guess to fix this

Hi :)

So this is it :

Sample LPCM (http://www.mediafire.com/download.php?nzmywyizynx)

I have more information (grrrrrrrrr) it didn't see yet my AVR (maybe tomorrow) and after it will repair !

Big thx :)

[albain]

Thanks, got it

[whurlston]

albain, I've asked a few people with NVidia HDMI cards (like the 2xx series) to test DD+ bitstreaming. When I had mine, DD+ was listed as a supported compressed format in the Windows Audio Device properties.

[albain]

Okay

About LPCM again, the stream outputted by FFDShow is completely different than the stream outputted by TMT.
Tomorrow I will try to understand why FFDShow does not go through the LPCM highdef code path.

[Andy o]

albain, I've asked a few people with NVidia HDMI cards (like the 2xx series) to test DD+ bitstreaming. When I had mine, DD+ was listed as a supported compressed format in the Windows Audio Device properties.

I just replied to your messsage, but I now remembered that I did check it out before (http://www.avsforum.com/avs-vb/showthread.php?p=17594615#post17594615). I'm gonna try again though with WinDVD and this ffdshow beta, but DD+ doesn't show in supported formats anyway, like with your 200 card.

[whurlston]

I just replied to your messsage, but I now remembered that I did check it out before (http://www.avsforum.com/avs-vb/showthread.php?p=17594615#post17594615). I'm gonna try again though with WinDVD and this ffdshow beta, but DD+ doesn't show in supported formats anyway, like with your 200 card.

Ah, if the supported format doesn't show up, it most likely won't work. I'm waiting to hear from a couple other people too.

[whurlston]

Disclamer: USE AT YOUR OWN RISK.

Download and install the revision 3149 x64 build from http://www.xvidvideo.ru/ffdshow-tryouts-project-x86-x64/ffdshow-tryouts-project-svn-3149-x86-x64.html

When installing, it wanted to install to c:\Windows\system32. I changed this to c:\program files\ffdshow keeping my 32 bit installation in c:\Program Files (x86)

Once installed, rename ffdshow.ax in the x64 install folder and repace it with the following:

Release version: http://www.epgtools.com/WilliamHurlston/ffdshow/ffdshow_x64_v3149_bitstream.zip
Debug version (if anyone wants it): http://www.epgtools.com/WilliamHurlston/ffdshow/ffdshow_x64_v3149_bitstream_debug.zip

Go to Start->All Programs->ffdshow64->Audio Configuration
In the "Codecs" section, set the following to "libavcodec":

AC3
E-AC3
TrueHD
MLP
DTS

In the "Output" section check passthrough for 16 bit and the HD Audio Formats.

You can use DSGraphEdit or GraphStudio64 to test 64bit graphs.

[mrcorbo]

@ whurlston

Do you think this could work in W7MC, or is W7MC locked to WMF for .m2ts files?

Things would get very interesting if you could launch HD movies in full video AND audio fidelity from within W7MC using My Movies or the like.

[whurlston]

@ whurlston

Do you think this could work in W7MC, or is W7MC locked to WMF for .m2ts files?

Things would get very interesting if you could launch HD movies in full video AND audio fidelity from within W7MC using My Movies or the like.

Yes. The same rules apply as x86 WMC though to override the Microsoft decoders. And you will need the x64 build on the MPC-HC standalone splitter. I good step by step by renethx is at http://www.avsforum.com/avs-vb/showthread.php?p=16951614#post16951614

[piit]

First off, you absolutely want to test on build 49.

Well, guess what - dts-HD HR works perfect with build 49 :-) Great work.

I've got another small issue now - this time with TrueHD. When I put the computer to sleep and resume, TrueHD passthrough (isn't passthrough better than bitstreaming? ;) ) wouldn't work - ffdshow reverts to decoding to PCM. (The screen flickers green/black a couple of times at the beginning - presumably when ffdshow reconnects to try different format). dts-HD MA bitstreams fine all the time.

Here's the log when TrueHD bitstreaming fails: http://pastebin.com/f3a21a075
Hope it helps.

My setup: ATI 5750, Denon AVR-1910, ffdshow bitsHD beta 49.

By the way, albain, are you planning to commit your HD audio bitstreaming changes to the ffdshow tryouts trunk any time soon? Or are you holding off until LPCM passthru is resolved?

Thanks for your great work, guys! If there's any way I could help I'd love to - please let me know.

Pete

[rica]

Pete, you are welcome to the board. :)
But i can't differ what you mean with (isn't passthrough better than bitstreaming?)
Is it an word trick?
While passtru= bitstreaming?
On the other hand, bitperfect streaming is different as you know :)

[madshi]

Huh, a new thread!

@Madshi : if you are around you may have some ideas about why a LPCM stream (DVD audio) should be reformat to LPCM (24 bits) in output, it should be a matter of passthrough ?
I think this is the difference between small and big endian. Usually decoders give you small endian, while Blu-Ray LPCM tracks are big endian. Audio renderers expect normal audio data to be small endian, so Blu-Ray LPCM tracks need to be converted, if you want them to behave similar to decoded audio data.

I suspect that LPCM streams need to be encapsulated in IEC headers as for AC3/DTS/TrueHD/DTSHD/EAC3 streams.

EDIT : actually, the header is different. I have the docs but there seems to be some bunch of work
As far as I can see, there are 2 *very* different ways to handle LPCM. First of all, LPCM is not different to the output of any decoder (FLAC, AAC, MP3, WMA, whatever), except that Blu-Ray LPCM is big endian instead of small endian, and maybe the channel sorting is different. However, the big difference is in how you send the data to the audio renderer. You can either:

(1) Send the audio data as decoded PCM samples to the audio renderer. This is how everyone has been doing things for ages. This is how ffdshow is doing things right now. The result of this method is that both the OS and the audio driver may do processing on the audio data. Basically what you feed the audio renderer will in most cases be very much different to what actually gets output to HDMI (unless you use wasapi exclusive mode, maybe).

(2) Send the audio data encapsulated as "LPCM bitstream". I don't know if this is possible at all. I hope it is. If it's possible, the audio data output through HDMI should be 100% identical to what you feed the audio renderer. This is what we're looking for. You can easily test whether it works by checking whether the OS volume control has any effect. If the volume control has an effect, the audio renderer doesn't treat the audio data as LPCM bitstream, but it treats it as normal decoded audio data instead.

Of course it would be *great* if you could get (2) to work. Because that would mean that we wouldn't even need wasapi exclusive mode, anymore. I have no idea how the media type information and audio data needs to be formatted for the audio renderer and audio drivers to understand the data as "LPCM bitstream". It's quite possible that you have to convert the audio data to big endian and use IEC headers etc. If you could get this to work, it should also work to passthrough the decoding result of FLAC, AAC, WMA etc decoders untouched via HDMI to the receiver...

P.S: I don't think you should add special code to ffdshow to passthrough LPCM input. Instead I think you should (a) convert LPCM input to normal PCM, which ffdshow is probably already doing right now. And then you should (b) offer an option to convert *ANY* PCM data to "LPCM bitstream" for HDMI transport. Only this way the whole solution would also work for losslessly transporting untouched FLAC, AAC, WMA etc decoding results...

[Skinleech]

I watched my first full title, bitstreaming TrueHD last night. For the most part, it was perfect, though I did suffering from occaisional micro-stutter, and one full audiop drop where my receiver lost the signal, but picked it back up straight away again.

I don't know if this is a system issue, Ffdshow issue or Xonar issue, I assume system as it seems to be working OK for others. I plan on doing a reinstall tonight (Win 7 this time as I'd prefer to be running that over Vista SP1) so will update on how that works out once done. Hopefully I'll be able to report successul results. Logs will follow if not.

[piit]

Pete, you are welcome to the board. :)
But i can't differ what you mean with (isn't passthrough better than bitstreaming?)
Is it an word trick?
While passtru= bitstreaming?
On the other hand, bitperfect streaming is different as you know :)

Thanks, rica :) As I said I'd love to take part in the development process in some way. Although I'm not experienced C++/DirectShow programmer I know fair bit about programming and I hope I could sometimes be helpful at least with testing and stuff like that :)

By saying "isn't passthrough better" I meant isn't it a better term for what we're doing? :) It is the same thing (as you noted), so why invent another word for it? :) Passthrough is IMHO much more descriptive, it says what is actually happening - AC3/DTS/TrueHD is passing through the computer and soundcard to the AVR untouched. But never mind, bitstreaming seems to be an established term by now. Unfortunately :)

By the way, is anybody experiencing the same issue - TrueHD passthrough not working after resuming from S3 sleep? Or is it just me? :)

Cheers
Pete

[piit]

I watched my first full title, bitstreaming TrueHD last night. For the most part, it was perfect, though I did suffering from occaisional micro-stutter...

Well, couldn't this be attributed to video synchronization issues? Unless you are running a 24 fps video (and I mean 24, not 23.976) with 24 (or 48) Hz screen refresh, there will always be synchronization issues. Only found out about this recently - I quite wish I didn't now! :) If only there was a way to synchronize the screen refresh to the video, not the other way round... :) The only way to have perfectly smooth Blu-ray video without any stutter whatsoever now is to render sound through ReClock - but that means no DTS-HD/TrueHD passthrough (because ReClock needs to resample the audio, unless you have a bloody expensive soundcard that allows setting its clock to an arbitrary value and therefore output bit perfect PCM at an arbitrary samplerate - anyway, the output needs to be PCM).

Sorry, gone a bit OT :) We have watched Taken with DTS-HD MA on Saturday and it was very good, without any issues and no A/V sync issues even after multiple seeks.

Pete

[rica]

Great ! You got the structures in it : LPCM is sent as IEEE_FLOAT type (which is PCM 32 bits float)
Could you also make those tests on Vista for DTSHD and TrueHD with PDVD9 ?
This will tell us which structures to feed for bitstream support on Vista, which is the missing key to finish the job.
Let's continue this discussion on the other thread.
This is a very good news !:)

By pleasure :)

When i back this evening.

[blackEyEz]

Seems like christmas is early this year :) TrueHD and DTS-HD bitstreaming!! Is there already an early estimate when a final version will be released or will you guys wait till you got LPCM and DD+ working?

And when a final version comes out can i use MediaPortal for HD bitstreaming? Or do they need to an update too?

Awesome work guys!

[madshi]

Great ! You got the structures in it : LPCM is sent as IEEE_FLOAT type (which is PCM 32 bits float)
Well, that's weird. Does HDMI support 32bit float LPCM!? I thought it only supported 16bit and 24bit cardinal LPCM?

[Skinleech]

Well, couldn't this be attributed to video synchronization issues? Unless you are running a 24 fps video (and I mean 24, not 23.976) with 24 (or 48) Hz screen refresh, there will always be synchronization issues. Only found out about this recently - I quite wish I didn't now! :) If only there was a way to synchronize the screen refresh to the video, not the other way round... :) The only way to have perfectly smooth Blu-ray video without any stutter whatsoever now is to render sound through ReClock - but that means no DTS-HD/TrueHD passthrough (because ReClock needs to resample the audio, unless you have a bloody expensive soundcard that allows setting its clock to an arbitrary value and therefore output bit perfect PCM at an arbitrary samplerate - anyway, the output needs to be PCM).

Sorry, gone a bit OT :) We have watched Taken with DTS-HD MA on Saturday and it was very good, without any issues and no A/V sync issues even after multiple seeks.

Pete

Perhaps. But I am running a Powerstripped 23.976 display with 23.976 material. It doesn't happen with TMT3 from original disk or folder structures, and not does it happen when bitstreaming ac3 or DTS using ffdshow or any other directshow sp/dif capable decoder.

Maybe it is more sensitive as I do get some dropped frames, but as noted, the audio never drops on other material.

[piit]

Perhaps. But I am running a Powerstripped 23.976 display with 23.976 material.

Well, that's another story! Sorry, didn't realize you're using Xonar - ATI 5xxx unfortunately isn't supported by PowerStrip yet :( Lucky you :) This looks like the last hurdle before one card rules them all :rolleyes:

[albain]

Well, that's weird. Does HDMI support 32bit float LPCM!? I thought it only supported 16bit and 24bit cardinal LPCM?

I don't think so and I am not sure that PowerDVD outputs LPCM but PCM 32 bits float

However I am sure that FFDShow does. Does this code sounds correct to you ? This code is applied only if high definition LPCM 24 bits is detected. I am not very familiar with big endian formatting but it looks like it ?

size_t end1 = (src.size()/12)*12;
for (size_t i = 0 ; i < end1 ; i += 12) {
*dst++ = (src[i]<<24) + (src[i+ 1]<<16) + (src[i+ 2]<<8);
*dst++ = (src[i+3]<<24) + (src[i+ 4]<<16) + (src[i+ 5]<<8);
*dst++ = (src[i+6]<<24) + (src[i+ 7]<<16) + (src[i+ 8]<<8);
*dst++ = (src[i+9]<<24) + (src[i+10]<<16) + (src[i+11]<<8);
}
size_t end2 = (src.size()/3)*3;
for (size_t i = end1 ; i < end2 ; i += 3)
{
*dst++ = (src[i ]<<24) + (src[i+ 1]<<16) + (src[i+ 2]<<8);
}

[madshi]

I don't think so and I am not sure that PowerDVD outputs LPCM but PCM 32 bits float

However I am sure that FFDShow does. Does this code sounds correct to you ? This code is applied only if high definition LPCM 24 bits is detected. I am not very familiar with big endian formatting but it looks like it ?
Have you read my other post (no 168)?

Yes, that ffdshow code looks like it converts big endian 24bit to small endian 32bit (or small->big). But both cardinal, not float.

[albain]

Yes I have but I misunderstood how FFDShow works

Actually, LPCM (even 24 bits highdef) streams are converted to PCM 32 bits in order to be processed by FFDShow filters.

However, the output filter is able to format the stream to the following :
- AC3 encoding
- PCM 16,24,32,32 float
- LPCM 16 bits

But not LPCM 24 bits. This code I showed you is actually as you pointed out not a conversion to LPCM but a conversion from LPCM 24 bits to PCM 32 bits (cardinal).

And also PowerDVD won't output LPCM is because from the dumper (don't have the buffers content though) the media structure sent to the renderer is "32 bits float".

So I don't know if anyone already saw 96Khz/24 bits/7.1 LPCM on his receiver from a software player.

EDIT :
Anyway, it should not be very complicated to do the reverse conversion from PCM to LPCM 24 but it would be like make a passthrough on LPCM streams (and conversion on PCM) before doing that I want to be sure if it will be accepted by the renderer

FYI, LPCM 16 outputted by FFDShow sends the following mediatype to the renderer : MEDIASUBTYPE_DVD_LPCM_AUDIO
Maybe it could be used for LPCM 24 too (by changing the values of the bps...)

[madshi]

Actually, LPCM (even 24 bits highdef) streams are converted to PCM 32 bits in order to be processed by FFDShow filters.

However, the output filter is able to format the stream to the following :
- AC3 encoding
- PCM 16,24,32,32 float
- LPCM 16 bits

But not LPCM 24 bits. This code I showed you is actually as you pointed out not a conversion to LPCM but a conversion from LPCM 24 bits to PCM 32 bits (cardinal).
That makes a lot of sense.

Anyway, it should not be very complicated to do the reverse conversion from PCM to LPCM 24 but it would be like make a passthrough on LPCM streams (and conversion on PCM) before doing that I want to be sure if it will be accepted by the renderer
That's fine with me, but is the original LPCM data structure as stored on Blu-Ray the same thing that HDMI wants? I'm not sure about that.

FYI, LPCM 16 outputted by FFDShow sends the following mediatype to the renderer : MEDIASUBTYPE_DVD_LPCM_AUDIO
Maybe it could be used for LPCM 24 too (by changing the values of the bps...)
I didn't even know there was an option for "LPCM 16". Do we know whether the OS and audio drivers leave "LPCM 16" data alone or whether they process it? E.g. does the OS volume control affect "LPCM 16" ffdshow output?

[albain]

It is said in FFDShow comments that this option could be used to connect to Cyberlink DVD audio filters

I believe that this is functional but nobody seems to use it or knows about it.

And I guess that you are right about LPCM encoding through HDMI.
In the docs I supplied to you there is a dedicated doc on LPCM formatting (doc #3).

I guess I'll have to get into it but I'd rather have dumper logs too

[SamuriHL]

Well, that's weird. Does HDMI support 32bit float LPCM!? I thought it only supported 16bit and 24bit cardinal LPCM?

PDVD9 is sending 32 bit for LPCM 48/24 or higher.

[rica]

Do we know whether the OS and audio drivers leave "LPCM 16" data alone or whether they process it? E.g. does the OS volume control affect "LPCM 16" ffdshow output?

Good point; i will check it for PDVD as weel :)

[albain]

Is there an existing software player outputting LPCM ?

[Sebastiii]

Hi
Good question :)
:)

[albain]

Otherwise, someone asked a question about when the HD audio branch will be committed to the trunk

The answer is quite simple : once HD audio bitstream will work on Vista.

For that I am waiting for PDVD9 dumper logs on Vista to get the right media structures. Rica is on his way to provide them

About LPCM, there is in my opinion too much work to make it work so this will be done later after the merge.

[Sebastiii]

Sorry that i can't help for now :(
But when i log on TMT and Xonar before it was on Vista, so why it doesn't work ? i'm lost :) lol

[albain]

Sorry that i can't help for now :(
But when i log on TMT and Xonar before it was on Vista, so why it doesn't work ? i'm lost :) lol

No problem, let's say a little prayer for your AVR :-)

[Sebastiii]

Yes, yes :) i hope with all my finger :)

[whurlston]

Is there an existing software player outputting LPCM ?

PDVD 9 does. Playing a 96/24 track now. Interestingly, it does not output LPCM 48/16 but runs it through Windows Mixer. Anything higher, it outputs the LPCM though.

[Browncoat]

I am waiting for PDVD9 dumper logs on Vista to get the right media structures.
Don't think it is helpfull, just in case. PDVD9+dumper with Auzentech on Vista x86 (before crash).

DTS HD MA
http://www.mediafire.com/?n1iey4ufmxy

Dolby TrueHD
http://www.mediafire.com/?jizzz5gedhj

Don't forget about Auzen :)

[albain]

Don't think it is helpfull, just in case. PDVD9+dumper with Auzentech on Vista x86 (before crash).

DTS HD MA
http://www.mediafire.com/?n1iey4ufmxy

Dolby TrueHD
http://www.mediafire.com/?jizzz5gedhj

Don't forget about Auzen :)

The last media type sent is WAVE_FORMAT_DOLBY_AC3_SPDIF.
Maybe (and probably) this is not this one which is used but this is the last one I can see.

And with this you get HD audio bitstream ?

Normally if PDVD has updated with the new patch his media types for the ATI, auzentech should have updated his drivers too ?

[Browncoat]

And with this you get HD audio bitstream ?
Not with dumper (crash...). Without dumper everything OK.

auzentech should have updated his drivers too ?
There is no driver update for Auzen. AFAIK there are no problems with Auzent+PDVD last patch. I'll check that by myself later.

EDIT
Now i have PDVD build 2320. Bitstreaming works well on Seven x64. Logs from Vista x86 will be tomorrow (i don't think there will be any difference). It's time to go to bed now ))

[rica]

OK albain and friends.

I'm here but i'm afraid i don't have good news for you.

I found those forced debugs are useless; because:

I have to open PDVD to see its PID which changes in every restarting of the application. (needed for dpinpurr)

When i mount an ISO, PDVD starts and "loading BD" slidebar stops at 98% and application crashes.

And i need to see the settings page of PDVD so as to check HD audio bitstreaming option everytime (or at least open that page and click to "OK") to get DTSHD or THD bitstreaming otherwise PDVD, at the beginning, starts decoding to PCM as default. (and unfortunately crash happens before the possibility of setup bitstreaming option.)

But anywway (even i don't thrust them) i add this DTSHDMA bitstreaming??? debug log on Vista:

http://www.mediafire.com/?t4mmjhmiiuj

and a THD bitstreaming???? debug log on Vista:

http://www.mediafire.com/?tdwlktwy0my

BTW, some more info:

LPCM 16 bit never bitstreams on Vista or 7 with PDVD , you can check it with volume slide bar; volume always changes when you move it which means LPCM is decoded to PCM by Cyberlink decoder in PDVD and it's not a bitstream at all.

(volume slide bar or mixer settings never effect anything under the cases:

1) Kernel(XP) or WASAPI (Vista) bitperfect streaming_decoded audio in SW but goes directly to your AVR bypasssing MS mixer.

Or,

2) Bitstream _non-decoded compressed audio directly goes to your AVR bypassing MS mixer.)

While debugging ffdshow we get the right debug log since we open dbgview at the beginning and play ffdshow without any crash.
We need to fix this while debugging PDVD.


The result of my trials for now: (over Ati 5*** series)

ffdshow and PDVD9 bitstream THD, DTSHD-MA, DTSHD-HR very well. (on Vista and 7)
ffdshow and PDVD9 can not bitstream LPCM for now. (on Vista and 7)

EDIT:


LPCM 24/48 bitstreaming works via PDVD9; given it a go with Vista and 7/64 while 16 bit doesn't work in any case.
24 or 16 bit LPCM bitstreams dedected as DD by ffdshow for the time being.
I can not give any true debug logs since PDVD application crashes at the beginning.


_ _ _ _ _ _ _

_ _ _ _ _

[SamuriHL]

That doesn't match my results. 48/16 always goes through the windows audio stuff, but, 48/24 with the non-decoded HD audio option bypasses the windows audio stuff and is for all intents and purposes bitstreamed. When I check the info screen, it's showing that a 48/24 track is being output as 32 bit.

[rica]

That doesn't match my results. 48/16 always goes through the windows audio stuff, but, 48/24 with the non-decoded HD audio option bypasses the windows audio stuff and is for all intents and purposes bitstreamed. When I check the info screen, it's showing that a 48/24 track is being output as 32 bit.

The problem, all LPCM BDs are 48/16 which i belong to.
If you upload a 48/24 LPCM i'll be appreciated.

[SamuriHL]

I'll see what I can do. It definitely won't be tonight. I've been up to my neck in work the past couple days. If someone else has a 48/24 or higher LPCM track that'd be useful. Note that it MUST be in ISO form so that it can be mounted. Rica, I'm sure you know how to take an m2ts or mkv and convert it to ISO, right?

[rica]

Note that it MUST be in ISO form so that it can be mounted. Rica, I'm sure you know how to take an m2ts or mkv and convert it to ISO, right?

:D:D:D

Thanks :)

[rica]

samuri, are we talking about Ati passthrough? Or Xonar?

And anybody?

If someone else has a 48/24 or higher LPCM track that'd be useful.

[whurlston]

samuri, are we talking about Ati passthrough? Or Xonar?

And anybody?
ATI Passthrough.

That doesn't match my results. 48/16 always goes through the windows audio stuff, but, 48/24 with the non-decoded HD audio option bypasses the windows audio stuff and is for all intents and purposes bitstreamed. When I check the info screen, it's showing that a 48/24 track is being output as 32 bit.

That's my results as well. Renethx has also confirmed this on avsforum. One thing that I noticed though is that even though the PDVD Info screen shows 32 bits, when I connect to the graph and view the pin configuration,, it reports 24 bit (this is the report from a 96/24 2.0 LPCM track):

20. [ 0006]/(Audio Out) -> [CyberLink Audio Renderer (PDVD9) [AR1]]/(Audio Input pin (rendered))
Major: MEDIATYPE_Audio
Subtype: {6AD2A952-9979-4DC4-B6E1-6F3C3137B342}
bFixedSizeSamples: FALSE
bTemporalCompression: FALSE
lSampleSize: 1
cbFormat: 18
Format: FORMAT_WaveFormatEx
WAVEFORMATEX:
wFormatTag: 0xc000 (49152)
nChannels: 2
nSamplesPerSec: 96000
nAvgBytesPerSec: 576000
nBlockAlign: 6
wBitsPersample: 24
cbSize: 0