Audio enhancements

[peteru]

I've done some work on improving the audio capabilities. At this stage, all works is done on the T4 first, but the intent is to port as much as possible to the T2/T3 and U4.

For those of you who are interested in testing an alpha version of the enhanced T4 driver, back up your settings, make a copy of the old driver, download the new driver, install it and reboot. If the driver doesn't work well enough for you, you can restore the old driver and reboot. If the new driver results in an unbootable system (and it should not), reflash from USB and restore settings.

An alpha snapshot of the T4 driver can be downloaded from here: ini-840t4-drivers-3.14.2-20180220.zip

Code: Select all

root@beyonwizt4:~# wget https://peteru.net/ini-840t4-drivers-3.14.2-20180220.zip
root@beyonwizt4:~# unzip ini-840t4-drivers-3.14.2-20180220.zip
root@beyonwizt4:~# cp /lib/modules/3.14.2/extra/dvb.ko dvb.ko.original
root@beyonwizt4:~# cp dvb.ko /lib/modules/3.14.2/extra/dvb.ko

The changes are:

• Add HDMI identification of "Beyonwiz T4", vendor "Beyonwiz"
  On receivers that can display HDMI source device identification, the correct device label will now appear.
• Add missing capabilities for AUDIO_GET_CAPABILITIES
  When advertising driver capabilities to user space, advertise all supported formats.
• Fix playback of AAC files
  Some AAC files did not play at all or took many seconds to start playing. In some instances, playing back AAC-HE files could lead to hardware lockups or system crashes.
• Ensure there are enough audio resources to handle all codecs
• Enable AC3+ and AACSBR audio passhtrough
• Downmix to 5.1 and enable high bit rate passthrough over HDMI

The last point on the list requires a bit more explanation. It's the big one.

In previous versions of the driver, audio could be delivered in only two formats. It could either be decoded and downmixed to stereo or passed through as a standard bit rate bit stream. The analogue outputs are hooked up to a stereo DAC, so those outputs must always receive the decoded and downmixed stereo signal. The SPDIF output can receive either downmixed stereo or act as passthrough for any standard bitrate codec. The SPDIF output can not handle any high bit rate codecs or more channels than stereo when sending decoded audio. In the old drivers, the HDMI port output was configured to receive a copy of whatever was being sent out to SPDIF. That means that the HDMI audio output had all the limitations of SPDIF.

In the new drivers, the HDMI port and SPDIF ports have been decoupled and the pipelines have been enhanced so that HDMI audio output can deliver a much better experience. When using down mix mode with the new drivers, audio will be decoded to 5.1 multichannel format. For DAC and SPDIF output, the audio will be downmixed to stereo. HDMI output will receive 5.1 multichannel decoded signal. In passthrough mode, the codecs will be divided into two categories, standard and high bit rate. DAC output will always receive decoded audio downmixed to stereo. SPDIF will receive passthrough bit streams for standard bit rate codecs and downmixed stereo signal for high bit rate codecs. HDMI will receive both standard and high bit rate codec passthrough bit streams.

Examples of standard bit rate codecs are AC3 and DTS. Examples of high bit rate codecs are AC3+ and DTS-HD.

This is still work in progress, there are lots and lots of combinations to try. Some things are likely to be broken, but probably less broken than they were with the old drivers. If you feel like testing, please do so and let me know your results. I am interested in both positive and negative feedback. It will be good to know what was tested and works, so that if there are regressions in the future, we can track down the changes.

[glen]

Hi Peteru

Wow, some of these enhancements sound (no pun intended) really good. Thank you. Cannot wait until they are ported to the U4.

Glen

[peteru]

There appears to be virtually no gear out there that can decode AAC bit streams over SPDIF or HDMI. Even if there are devices out there that can do it, I don't have anything to test with.

I'm considering just ditching the AAC passthrough option and always decoding AAC. HDMI will receive a 5.1 channel PCM downmix (or 7.1, if I can enable that in the future), SPDIF and DAC will receive stereo downmix.

Any strong objections? If you object, please explain!

[MrQuade]

I'm considering just ditching the AAC passthrough option and always decoding AAC. HDMI will receive a 5.1 channel PCM downmix (or 7.1, if I can enable that in the future), SPDIF and DAC will receive stereo downmix.

Any strong objections? If you object, please explain!

Nope, that sounds perfect. I have seen several devices that claim to be able to send an AAC bitstream, but am yet to find ANYTHING that can actually accept/decode it.

Software such as Kodi deal with this by "downmixing" to Dolby 5.1, which while acceptable for SPDIF due to bandwidth limitations, is not ideal for HDMI connections due to the lossy nature of the downmixing. I would much prefer a 5.1 PCM downmix. (BTW, This was just for illustration of an approach taken by others, I am not suggesting that it is a good idea for the T4 to downmix AAC5.1->DD5.1)

[IanSav]

Hi PeterU,

There appears to be virtually no gear out there that can decode AAC bit streams over SPDIF or HDMI. Even if there are devices out there that can do it, I don't have anything to test with.

I'm considering just ditching the AAC passthrough option and always decoding AAC. HDMI will receive a 5.1 channel PCM downmix (or 7.1, if I can enable that in the future), SPDIF and DAC will receive stereo downmix.

Any strong objections? If you object, please explain!

If leaving the code in place doesn't "cost" anything then I would suggest leaving it available.

I understand that you don't have anything with which to test the feature now but that doesn't mean that such equipment does not, or will not, exist. If the passthrough feature is removed then there will be no way for you to collect feedback from users who have, or obtain, compatible equipment.

Regards,
Ian.

[MrQuade]

I understand that you don't have anything with which to test the feature now but that doesn't mean that such equipment does not, or will not, exist. If the passthrough feature is removed then there will be no way for you to collect feedback from users who have, or obtain, compatible equipment.

I am going to just put it out there and predict that there will never be any consumer gear that supports it. AAC has been around for 20 years, and I know of equipment that has supported HDMI bitstreaming output for more than 10, but still nothing can receive it. If it hasn't happened by now, it is not going to.

The feature has tripped up several users, and is just confusing to leave it in the T series menus. I did notice that the U4 doesn't include the option, so I don't think it needs to be added in.

PCM output over HDMI achieves everything that could be achieved with AAC bitstreaming, and any new equipment is very unlikely to be limited to SPDIF audio-only.

[IanSav]

Hi MrQuade,

While its use may not be prevalent it is in the standards so it should not be dismissed trivially.

Regards,
Ian.

[raymondjpg]

Software such as Kodi deal with this by "downmixing" to Dolby 5.1, which while acceptable for SPDIF due to bandwidth limitations, is not ideal for HDMI connections due to the lossy nature of the downmixing.

Are you sure about that? I was under the impression that Kodi could transcode AAC, but only to two channel AC3, not multi-channel.

I would much prefer a 5.1 PCM downmix. (BTW, This was just for illustration of an approach taken by others, I am not suggesting that it is a good idea for the T4 to downmix AAC5.1->DD5.1)

Are we talking about downmixing or transcoding? I for one would dearly like to see any piece of hardware capable of transcoding multi-channel AAC to AC3. I have nothing against quality of sound from PCM, rather that AV receivers in my opinion make better use of a stereo AC3 stream in their various surround formats, for example Dolby Prologic II Movie in Yamaha receivers.

I am going to just put it out there and predict that there will never be any consumer gear that supports it. AAC has been around for 20 years, and I know of equipment that has supported HDMI bitstreaming output for more than 10, but still nothing can receive it. If it hasn't happened by now, it is not going to.

Never say never, but I've been waiting waiting as well and given up.

PCM output over HDMI achieves everything that could be achieved with AAC bitstreaming, and any new equipment is very unlikely to be limited to SPDIF audio-only.

Well, not quite, as I alluded to previously. For that reason with my HTPCs I decode AAC to PCM, then transcode on the fly to AC3.

[MrQuade]

Are you sure about that? I was under the impression that Kodi could transcode AAC, but only to two channel AC3, not multi-channel.

That's not what I said. I said Kodi downmixes aac5.1 to dolby5.1.
I was just using the same terminology that PeterU used in his post, but yes, I would normally call that transcoding.

Well, not quite, as I alluded to previously. For that reason with my HTPCs I decode AAC to PCM, then transcode on the fly to AC3.

I think you are mistaken on that point.
A native Dolby track will be better than a Dolby track that has been converted to PCM because native Dolby has a bunch of extra metadata that your receiver can make use of.

A aac track converted to Dolby won't have any of that metadata so there is no benefit in your receiver.
In fact, since the aac to Dolby conversion requires recompression, you are going to get a loss of audio quality. You are far better off going from aac to PCM and leaving it at that since the conversion is lossless.

[MrQuade]

Will I can confirm that DD+ both bitstreams and transcodes to multichannel PCM.

AAC 5.1 is also transcoding to 5.1 PCM too

I haven't given them more than a quick check, so not sure how they go for extended periods.

Re-reading your notes, I realise now that you mentioned this, bit I didn't parse it. Something I noticed when testing, my stereo AAC material was also covered to 5.1 PCM. I tried a downloaded a media file and the Racing channel (aac-he). I think that those should be getting converted to PCM 2.0 so that the receiver can make the correct decisions as to how the audio is mapped to the available speakers.

[peteru]

I'm considering just ditching the AAC passthrough option and always decoding AAC.

If leaving the code in place doesn't "cost" anything then I would suggest leaving it available.

The existing code is clearly untested and looking at it, it probably doesn't work. The "cost" of keeping the code is code rot, unnecessary complexity, untested code paths, increased attack surface, binary and API bloat and a whole other host of software engineering do nots.

I understand that you don't have anything with which to test the feature now but that doesn't mean that such equipment does not, or will not, exist. If the passthrough feature is removed then there will be no way for you to collect feedback from users who have, or obtain, compatible equipment.

And there lies the rub. So far, no one has been able to identify any equipment that is capable of receiving and decoding any form of AAC over HDMI. Not multi channel, not even stereo. For all intents and purposes, such equipment does not exist and judging by the feature development in A/V devices over the last decade, it is unlikely to ever make an appearance. AAC decode support in the wild is limited to stereo decoded from files streamed over DLNA/HTTP or files read from USB/SMB volumes. Some equipment also supports stereo AAC over Bluetooth. Even the high-end $5,000 receivers have those limitations.

As far as I can see, there is no justification for keeping the AAC passthrough option. The only time that AAC passthrough comes up for discussion is because someone enables it and as a result they end up with no audio on some content.

If someone has equipment that can decode multi channel AAC over HDMI and the feature is working for them with their Beyonwiz, I'd be interested in hearing about it. My guess would be that the size of that group is zero.

[peteru]

While its use may not be prevalent it is in the standards so it should not be dismissed trivially.

Can you point me to specific sections of the relevant standards please? I'd like to check out the exact wording.

[peteru]

Are you sure about that? I was under the impression that Kodi could transcode AAC, but only to two channel AC3, not multi-channel.

That's not what I said. I said Kodi downmixes aac5.1 to dolby5.1.
I was just using the same terminology that PeterU used in his post, but yes, I would normally call that transcoding.

This is purely a nomenclature argument. I don't think MrQuade is using the same terminology that I would use, but I think we all understand what he is trying to convey.

At the moment the T4 driver can only pass through or decode (with an optional downmix) when it comes to the physical outputs. There is no encoder involved on this path. Transcoding (decoding, optional processing and then encoding) for streaming / recording is different story, which will have to be left for another day.

If Kodi can only decode two channel AAC and then encodes that as two channel AC3, then I'd class that functionality somewhere between pointless and a bad idea. Once you go to more than two channels and have a bit rate constrained interface (such as SPDIF), then using a multi-channel encoder for the output stage makes sense in some configurations. HDMI output is not one of those.

I think you are mistaken on that point.
A native Dolby track will be better than a Dolby track that has been converted to PCM because native Dolby has a bunch of extra metadata that your receiver can make use of.

Spot on. A quick listening test with the new driver will more than prove this, especially as you decrease volume from 0dB (reference level, i.e. too loud for normal use) to normal listening levels or even night time listening levels.

A aac track converted to Dolby won't have any of that metadata so there is no benefit in your receiver.
In fact, since the aac to Dolby conversion requires recompression, you are going to get a loss of audio quality. You are far better off going from aac to PCM and leaving it at that since the conversion is lossless.

Again, correct. In fact, receivers are more likely to do the right thing with multi channel PCM data than they are with "fake" AC3. When a receiver is fed AC3 stream, it will assume that there is certain metadata correctly mastered into the soundtrack. If that information is not correct, it will sound terrible. Watch any of the FTA broadcasters for a few hours and you'll be able to spot just how bad it can get.

my stereo AAC material was also covered to 5.1 PCM. I tried a downloaded a media file and the Racing channel (aac-he). I think that those should be getting converted to PCM 2.0 so that the receiver can make the correct decisions as to how the audio is mapped to the available speakers.

No such luxury in the SDK I am afraid. The code has to decide whether an uncompressed output will be stereo or multi-channel, initialise the device in that mode and that is the end of that. You don't get to arbitrarily mix or re-route channels or select the number of channels. At the moment multi-channel is defined as 5.1 (strictly speaking 6 channels), but I am investigating whether it is possible to enable 7.1 multi-channel mode.

Anyway, at this stage I'm not so much interested in arguing about the various potential features. There is enough complexity in there to keep entire teams of engineers in large companies like Yamaha and Denon busy for years on end. I'm constrained by the SDK, but have a fairly good appreciation for the most common use cases for the Beyonwiz boxes.

I am more interested in actual testing and reports on stability/reliability. It would also be good to know what works well. If there are things that do not work as well as you expect, it would be good to know the details. I'm aware of a couple of things, but I'll give it a few days to see if anyone picks them up. I don't want to spend too much time working on stuff that doesn't bother anyone.

[IanSav]

Hi PeterU,

Here are some references I have located:

• https://denon.custhelp.com/app/answers/ ... and-2.0%3F
• https://en.wikipedia.org/wiki/HDMI

AAC was added in HDMI 2.0. I can't seem to find any open access versions of the HDMI specifications. Apparently you need to be a member of the HDMI forum to get access to the documentation.

Here are some multi-channel AAC test files:
- https://www2.iis.fraunhofer.de/AAC/multichannel.html.

Regards,
Ian.

[MrQuade]

AAC was added in HDMI 2.0. I can't seem to find any open access versions of the HDMI specifications. Apparently you need to be a member of the HDMI forum to get access to the documentation.

That sounds like when specifically aac-he was added to the HDMI standard. AAC bitstreaming was something that the PS3 supported many years before that.

I think you are referring to something different.

[MrQuade]

Hmm, even the MPEG audio on SD channels is being converted to multichannel PCM.

This behaviour is really going to bugger with Pro-logic deciding if stereo sources.

I know that you aren't looking for feature requests, but if the T4 can't work the number of channels out itself, as you have indicated, then I think the user really needs a toggle to decide if the downmix is stereo or multichannel.

[peteru]

Hmm, even the MPEG audio on SD channels is being converted to multichannel PCM.

This behaviour is really going to bugger with Pro-logic deciding if stereo sources.

Can you please be a lot more specific on how this is a problem and how to demonstrate / test for it?

From my observations here, the Yamaha RX-V683 is not doing anything I would be unhappy with.

[MrQuade]

Can you please be a lot more specific on how this is a problem and how to demonstrate / test for it?

From my observations here, the Yamaha RX-V683 is not doing anything I would be unhappy with.

So say I am receiving an SD channel that is being broadcast with MPEG audio in stereo, and the movie I am watching was mastered as a Dolby Pro-logic track.

If the T4 upmixes that stereo track to 5.1, then my receiver is getting two channels of audio and 3.1 channels of silence. I don't ever get any audio from my rear speakers.

If instead the T4 sends the audio in its original stereo format, then the receiver recognises that as a Dolby Pro-logic signal and decodes the centre and surround channels appropriately.

Even if the audio track wasnt mastered with Pro-logic, as long as the receiver is receiving a stereo signal, then the receiver will have a crack at producing surround audio. If I am receiving several channels of silence instead, the receiver just faithfully reproduces the silence on those channels.

[IanSav]

Hi MrQuade,

That sounds like when specifically aac-he was added to the HDMI standard. AAC bitstreaming was something that the PS3 supported many years before that.

I think you are referring to something different.

We *are* talking about when AAC was officially added to HDMI. That does not discount some devices using AAC before the standard was ratified and officially adopted.

Regards,
Ian.

[peteru]

That sounds like when specifically aac-he was added to the HDMI standard. AAC bitstreaming was something that the PS3 supported many years before that.

I think you are referring to something different.

We *are* talking about when AAC was officially added to HDMI. That does not discount some devices using AAC before the standard was ratified and officially adopted.

No, MrQuade is right. This specific addition is for AAC-HE, which in reality is about 3 or 4 more variants of AAC added to the previous AAC support that was based on the initial MPEG-4 definitions of AAC.

Of course, none of that changes the fact that just because there is a standardised definition for supporting an optional encapsulation (that no equipment actually uses), it does not mean that any/all equipment needs to support it. In this instance, we have the opposite of an industry standard. It's an option that the entire industry has rejected/ignored.

Ian, I do appreciate your efforts to try and convince me, but so far the case is not very compelling. Maybe you can try to think about this from the user interface point of view:

• Does the feature benefit / make sense for the majority of users?
  No.
• Will the removal of the feature make the product unusable in instances where it worked before?
  No.

Sure, it would be cool to have a product that supports every bell and whistle, is easy to use and reliable at all times. As a single person working on this in my own time, I just can't do it. I think the best move is to remove the (unused) AAC passthrough option and concentrate on all the other features that are actually used by the users. In fact, just arguing the point here has probably taken away a few hours of my time that I could have spent looking at the code required to get this feature ported to T2/T3/U4.

If it turns out at some point in the future that AAC passthrough is a hot new feature that has wide industry support and is needed, we can always look at which products it needs to be implemented in and do a proper job of implementing and testing it at that time. I don't think it'll ever come to that.

[IanSav]

Hi PeterU,

You asked me a question so I tried to provide you an answer. I also provided you with some test media with which to test.

I don't have access to any contemporary equipment to test for this. My Yamaha RX-A3000 is now very old and doesn't support any of the current standards. I believe AAC over HDMI is meant to be working in HDMI 2.0 and HDMI 2.1.

If your time is so tight that you can ill afford to discuss the matter then it would seem that leaving AAC alone would be the path of least effort.

If you want to pull the AAC code then that is your prerogative. Given your upgrades have been funded by the community I hope you will restore the feature should the need or desire arise in the community.

Regards,
Ian.

[peteru]

So say I am receiving an SD channel that is being broadcast with MPEG audio in stereo, and the movie I am watching was mastered as a Dolby Pro-logic track.

If the T4 upmixes that stereo track to 5.1, then my receiver is getting two channels of audio and 3.1 channels of silence. I don't ever get any audio from my rear speakers.

If instead the T4 sends the audio in its original stereo format, then the receiver recognises that as a Dolby Pro-logic signal and decodes the centre and surround channels appropriately.

Even if the audio track wasnt mastered with Pro-logic, as long as the receiver is receiving a stereo signal, then the receiver will have a crack at producing surround audio. If I am receiving several channels of silence instead, the receiver just faithfully reproduces the silence on those channels.

Hah!

Now we are getting into the territory of different receivers behaving differently (or having different feature sets) when it comes to signal processing. I'm beginning to feel the urge to add another popular AVR to the development environment. Denon for the win, right? I'm only half joking - focused and persistent beta testers are also a valuable source of information.

The Yamaha isn't easily fooled. It has a surround processing auto-detect mode that can recognise just the above scenario and do the right thing. For example, it will take the 5.1 PCM input as decoded from broadcast MPEG and detect Dolby Prologic encoding on the L+R channels that actually have active data, then decode to the full 5.1 speaker config. You may want to explore your Denon and see if something similar is available.

At the same time, I'll study the SDK in more depth to see if there is anything I could do. I don't like my chances though. Even if there are hooks to fiddle with such things and IF it all actually works (that's a big IF based on my experiences so far), the number of possible combinations, together with the specific rules required to validate them easily go into the hundreds of thousands. Three physical outputs, times three input modes, times 50+ codecs, times approximately 8! channel assignments, times four downmix modes. Given enough constraints and assumptions, all this could probably be whittled down to a few hundred common combinations. Still too hard basket. Need to start thinking about this from a different angle...

So, low hanging fruit first, then refinements based on real-world experiences. Nothing wrong with blue-sky brainstorming, but only the best ideas with the best chance of success will make the short list.

[Paul_oz53]

Although I am sympathetic to the proposition that we should retain the option of aac, the discussion here persuades me that it is a feature of only theoretical value. It strikes me as a VHS vs Beta argument - ultimately both died, but only after one consumed the other.

To the extent I have a vote, I accept peteru's proposition that it be removed. Time is a precious commodity, especially where it is volunteered.

If a sudden but unexpected change in hardware were to occur, the resultant interest in the technology would drive the effort to adopt the capability. Until then, I don't see a reason to keep it.

My main gripe is retaining other Bwiz features that have no value to users now or in the forseeable future. Audio sync and offline decode - I'm looking at you.
Paul

[MrQuade]

The Yamaha isn't easily fooled. It has a surround processing auto-detect mode that can recognise just the above scenario and do the right thing. For example, it will take the 5.1 PCM input as decoded from broadcast MPEG and detect Dolby Prologic encoding on the L+R channels that actually have active data, then decode to the full 5.1 speaker config. You may want to explore your Denon and see if something similar is available.

Did you have a specific setting that controlled that in your amp[? I don't remember seeing anything like that in Denon land off the top of my head. I'll have a hunt about, but I don't like my chances.

[peteru]

The Yamaha isn't easily fooled. It has a surround processing auto-detect mode that can recognise just the above scenario and do the right thing. For example, it will take the 5.1 PCM input as decoded from broadcast MPEG and detect Dolby Prologic encoding on the L+R channels that actually have active data, then decode to the full 5.1 speaker config. You may want to explore your Denon and see if something similar is available.

Did you have a specific setting that controlled that in your amp

Have a look in the RX-V683 manual on page 61.

Although, now that I am trying to verify that it works as well as I said it did, I'm not having much luck. It could just be the selection of source material that I have available. Mostly a bunch of recordings from SBS dating back to 2016 or a selection of rage episodes going back a year.

[MrQuade]

Have a look in the RX-V683 manual on page 61.

Although, now that I am trying to verify that it works as well as I said it did, I'm not having much luck. It could just be the selection of source material that I have available. Mostly a bunch of recordings from SBS dating back to 2016 or a selection of rage episodes going back a year.

Thanks. Those look like the options in my X3400 (see manual here). I normally select the "Dolby Surround" option for stereo sources, and "Multi In + Dolby Surround" for multichannel PCM.

I tested a few of the options on a selection of stereo sources without success. Using my TV's tuner passing stereo, I veriifed that I could hear some surround activity, then switched to the T4 and timeshifted back over the same time period to compare and there was no activity from any speakers aside from the main fronts.

[peteru]

I tested a few of the options on a selection of stereo sources without success. Using my TV's tuner passing stereo, I veriifed that I could hear some surround activity, then switched to the T4 and timeshifted back over the same time period to compare and there was no activity from any speakers aside from the main fronts.

I think you may have been right all along and I was wrong about the Yamaha being able to figure out stereo encoded Dolby when fed 6 channel PCM. I think when I set it up initially, I had the speakers positioned such that the stereo imaging was so good that I may have been fooled into thinking there was a rear presence. I have since moved the speakers a few centimeters without redoing the calibration and it is now easier to localise each one. I have not heard anything from the rears.

I'll investigate the SDK options to see what, if anything, may be available.

[raymondjpg]

If the T4 upmixes that stereo track to 5.1, then my receiver is getting two channels of audio and 3.1 channels of silence. I don't ever get any audio from my rear speakers.

I do not have a T4 and my T2 is not connected to an AV receiver, so I cannot test this myself.

Is there an option to transcode an SD broadcast MPEG2 audio stream to AC3 before upmixing to 5.1, and how does the AV receiver cope with that?

[MrQuade]

Is there an option to transcode an SD broadcast MPEG2 audio stream to AC3 before upmixing to 5.1, and how does the AV receiver cope with that?

There is currently no option to do that, nor do I think you would want one.

As-per the Kodi discussion, you would be compromising audio quality with the re-compression to AC3.

If you were going to do any upmixing in the T4, you would simply upmix to PCM 5.1, but actually send audio data to all channels, rather than the current arrangement where 5.1 is sent with only 2 channels containing data. To properly upmix to 5.1, the T4 would need to decode Pro-Logic or DTS:X in software to perform the upmix. This would be functionally the same as your suggestion to upmix to AC3, but would avoid the loss of quality on recompression.

HOWEVER! You probably still wouldn't want to do it that way as it is generally preferable to have the receiver de-matrix the pro-logic encoded stereo signal, since the receiver is a specialised piece of audio hardware and presumably better at it. the receiver is also aware of how many speakers you have, and can upmix stereo to 5.1, 7.1, 9.2 etc etc as required.

The best solution possible is for the T4 to just pass on whatever audio it receives with as little processing as possible, and present it to the receiver as-such.

The only time transcoding to AC3 would really be of any use is if you were limited to an optical connection, and wanted to send multichannel aucio for an AAC 5.1 track. even then, you could argue that downmixing to a Dolby-Prologic-II PCM stereo stream may sound just as good/bad.

Also since PeterU said that the optical out here the same downmix as the analog outs, then you'd want that to be stereo anyway.

[raymondjpg]

There is currently no option to do that, nor do I think you would want one.

As-per the Kodi discussion, you would be compromising audio quality with the re-compression to AC3.

I'd say that you are straining at a gnat. We are not looking for audiophile quality sound here down a thin slice of wire in an HDMI cable.

[MrQuade]

I'd say that you are straining at a gnat. We are not looking for audiophile quality sound here down a thin slice of wire in an HDMI cable.

Perhaps, but AC3 offers no benefits over PCM in this scenario, so there is really no reason to use it.

[peteru]

Hmm, the inadvertent push of the alpha drivers into the beta feed has revealed that there is a lot of equipment out there that can not handle HDMI with multi-channel PCM audio properly, even though this has always been a mandatory part of the specification!

As a result a big chunk of work has surfaced - I'll need to implement HDMI audio format negotiation. I'll probably go quiet while I work on this. It's going to take a while to get something implemented and then probably many iterations to make it robust as people test with various combinations of equipment and report back issues.

After that, I'll return to the issue of changing the audio format on the fly, depending on what the codecs are producing. I'm not sure if that will be possible to auto-detect, it may need to be a user selectable option.

[peteru]

There is an online beta update available for the T4. With the following audio related improvements:

• HDMI hotplug video mode negotiation. Checks if the selected mode is supported by the HDMI sink and if it isn't, switches to the preferred mode advertised by the sink. That should put a stop to ending up with a video mode that can't be displayed. It also allows us to detect the number of audio channels supported by the connected device.
• Multi channel audio is now only sent to HDMI sinks that claim to support more than stereo. This should fix missing audio on HDMI devices that do not support multi channel PCM.
• Improvements to multi channel downmix. The codec is monitored for the number of channels being decoded. If the decoded content is stereo, the HDMI PCM output is switched to stereo, otherwise the HDMI PCM output is 5.1 PCM. This allows receivers to correctly apply surround decoding to stereo content. This feature can get interesting when watching commercial TV with Dolby encoded audio. As the content switch between various advertisements and programming, so does the encoding. This can result in dynamic switching between stereo and 5.1 PCM signal when downmixing.
• AAC downmix option has been removed. AAC is now always output as decoded PCM and obeys the above multi channel rules.

These changes had several hours of testing with my particular equipment and a whole host of media files. It seems robust enough in my configuration, but it'll be interesting to see how it behaves for others. Have a go and please report both positive and negative experiences.

[peteru]

In my investigations for possible codec additions, I've discovered that (at least on the T-series), it is not possible to support Dolby True-HD in either passthrough mode or to decode it.

On the other hand, I found a test clip that claims to have Auro3D information embedded in a DTS track and it plays OK. I can not be sure if the extra Auro3D metadata is being delivered to the receiver, the budget I had did not allow me to purchase an Auro3D capable unit. Even on the Denons, it is an extra option you have to purchase and it would require at least an 11.1 channel receiver and speakers to test properly.

[prl]

... at least an 11.1 channel receiver and speakers ...

[Henk]

On the other hand, I found a test clip that claims to have Auro3D information embedded in a DTS track and it plays OK. I can not be sure if the extra Auro3D metadata is being delivered to the receiver, the budget I had did not allow me to purchase an Auro3D capable unit. Even on the Denons, it is an extra option you have to purchase and it would require at least an 11.1 channel receiver and speakers to test properly.

http://www.avsforum.com/forum/90-receiv ... rsion.html

[peteru]

http://www.avsforum.com/forum/90-receiv ... rsion.html

Big thread, over 5,300 posts. Any particular posts in there that I should be looking at?

[Henk]

Big thread, over 5,300 posts. Any particular posts in there that I should be looking at?

Hi Peter,

More the time line (Sept 2014 to present) and the feeling that this particular development has not gone anywhere over this period of time, the people following it have pretty much given up on it.

With other words a dead branch in the evolutionary line....

Cheers,

Henk

[peteru]

OK, got it. If Auro3D doesn't quite work, there's no point spending time or money on making it work. On the other hand, if Auro3D happens to work for the few who can make use of the feature, great. So, provided "AS-IS" with no representations...

[peteru]

In other news, I uploaded T2 beta updates that bring some advanced audio features.

[MrQuade]

Brilliant work PeterU, thanks for the efforts. Being that multichannel stuff working must have been fun.

I haven't been able to test yet as I've been out fishing for the weekend but will do so asap.

A shame about the HD audio support, but at least DD+ is now in and that sound account for most people needs. Is there any chance of the base SD track of a DTS-HD track being supported?

Fingers crossed for the U4 anyway

[MrQuade]

Things seem to be mostly working as advertised so far. The DTS downmix is now working properly again, and the channel detection appears to be working for my 2.0 stereo AAC media.

I have run into an issue with at least two of my downloaded media files. They contain a 5.1 AAC audio track, but are only detected as stereo, and the T4 outputs them as such. I suspect it may be the result of a mistake made by whoever originally encoded these, and they have mistakenly marked them as stereo.

I normally wouldn't worry if it was a badly packaged media file, but I suspect that this kind of mis-marking may be commonplace, so it would be nice if the T4 could compensate.

MediaInfo reports attached.

EDIT: Also, maybe consider renaming the "Dolby digital / AC3 downmix" option to "Dolby digital / DTS downmix"?

[MrQuade]

Something has been bothering me a little about the automatic detection and implementation of multichannel PCM downmixing, and I was wondering if it would be valuable to have a setting that manually disables the multichannel downmix, and forces it to operate in stereo. The problem was, that I couldn't think of a case where it may be useful, given that the audio sink's capabilities are accurately reported.

Then it came to me.....Amps with extra Zones.
My Amp for example, supports a main zone, plus a zone 2 and zone 3. Zone 2 and 3 will only accept analog audio or PCM stereo as source audio formats. This is a very common limitation among receiver brands as I understand. It's because only the main zone can access the amp's primary audio processor.

If a user wanted to deliberately output stereo PCM audio so that their extra zone had audio, the amp would be advertising multichannel capability to the T4, which would then output multichannel PCM, but the amp would be unable to send the audio stream to its second zone.

I suggest that a new option be created called something like "Number of downmix channels" with valid values being "Auto" and "Stereo".

I realise that the particular use-case is possibly limited, but there may be other instances where this might be needed.

[peteru]

DTS-HD Master Audio and DTS:X already work on the T4. Both passthrough and downmix.

I'm yet to test on T2, but it should work.

[peteru]

the amp would be advertising multichannel capability to the T4

Would it? Surely if one of the HDMI inputs is already active and connected to the main zone, all the other HDMI inputs would correctly advertise a maximum of 2 channels. If the receiver advertises a capability that it can not support, I would consider the receiver HDMI implementation buggy and would be getting in touch with the AVR manufacturer regarding a firmware update. When you get a minute, test it out and see what happens.

One of the biggest problems with adding another downmix mode is that none of the enigma2 eco-system supports it. There is a lot of code in a lot of places that can only cope with what is essentially a boolean variable. Adding a third state to the mix will really screw things up.

Let's just see how well the current implementation works in practice and address any specific needs as they arise. Right now you have a hypothetical, atypical scenario.

[MrQuade]

Would it? Surely if one of the HDMI inputs is already active and connected to the main zone, all the other HDMI inputs would correctly advertise a maximum of 2 channels.
[/quote]
I am talking about the T4 being connected to an HDMI input on the receiver, and sending audio to the main zone, then the second zone is turned on and set to take audio from the same input.
When watching SD FTA, audio comes out of the second zone, and when switching to media with multichannel audio, the second zone sound was muted.

Let's just see how well the current implementation works in practice and address any specific needs as they arise. Right now you have a hypothetical, atypical scenario.

Well, technically not hypothetical, because I tested it and that's how it was.

I have no problem with that, but I did want to just throw it out there as a case where having a "force to stereo" would be desirable.

[MrQuade]

DTS-HD Master Audio and DTS:X already work on the T4. Both passthrough and downmix.

Hmm yes, you are correct of course.

The demo file I was using must have been an odd duck. I downloaded two other different demo files and they seem to work correctly.

In my bodgy demo file, there was a mutlichannel PCM track, an AC3 track and a DTS-MA track. The other two tracks played ok, but the DTS-MA did not (choppy audio), though it played properly on my PC and my HTPC (Kodi).

Probably nothing to worry about, but file the info away just in case.

[peteru]

Oh, I get it, you want to send the primary input to multiple zones, but your AVR can't do that with multi channel PCM. Does it have the same limitation for compressed audio, such as AC3? Kind of makes the multi-zone capability sound like a half-arsed idea thought up by one manufacturer and zealously copied by everyone else to tick the box on the feature comparison sheet. The Yamaha has a mode where it can specifically set zone 2 to copy whatever the main zone is playing (and I guess downmix to stereo as required), but that's only from consulting the manual. I have not actually tested.

Perhaps a parallel SPDIF connection for this purpose may work. I'll think about what can be done to address this need, but it's not at the top of the priority list.

[MrQuade]

No compressed audio world since the other zones can't access that part of the audio processor. Just analog and PCM stereo. And yes, an optical connection would be an effective workaround.

I was led to believe that is how most receivers handle their additional audio zones, but I haven't looked into it in great depth.