Yes you are right , probably audio over IP and no record at all.
Perhaps Stamatiski can give more detail about the audio pc processing he is doing and applications and workflow used.
+ Reply to Thread
Results 61 to 86 of 86
The purpose is streaming audio, only. Not even recording.
This is a photo to show you how I used to transmit HDMI audio up to now
[Attachment 51843 - Click to enlarge]
This is a 7.1 DAC which has been modified.
The audio is captured right before the analog section in i2S format. A i2s to SPDIF transceiver was giving the 8 LPCM channels in 4 SPDIF pairs.
But under the light of AoIP this is not enough for my needs anymore.
The OPPO 95 has two HDMI outputs and you can configure to use the one for video and the other for audio. Once the stream is received WITH identifiable windows drivers PER CHANNEL, then DANTE VIA should be able to translate them into DANTE protocol.
[Attachment 51841 - Click to enlarge]
Then the signal can be sent to virtually infinite destinations through this matrix DANTE controller
[Attachment 51842 - Click to enlarge]
I need this signal management to pass the signal simultaneously to various processors. My main DIY topic of interest has to do with active speakers. That is a whole other chapter beyond the scope of this thread so I would not want to add more irrelevant information
So, long story short: Once I can transmit successfully the 8 LPCM channels signal straight to my PC, I can stream it and MANAGE it.
I hope I gave you the picture.
Last edited by stamatiski; 4th Feb 2020 at 03:38.
Sorry to bother you again guys but could someone please post a screenshot of the Megawell drivers (deployed) as they appear on windows device manager and windows sound control panel?
Do they appear as separate channels or as single device?
Not sure if it's exactly what you do want but I made some screenshots from device manager, recording devices and audio options in Vdub2 related to Magewell pro Hdmi. The magewell has two audio capture pins and you need an application that can connect to the WDM that is the one that does multi channel (8) if the application do connect to the other pin you will have only audio down mixed to stereo 2 channels.
Edit: Add the two capture pins info zip file
Last edited by FLP437; 4th Feb 2020 at 15:13.
THANK YOU so much!
I just wanted to understand how windows acomodate the device. I believe that it will do the job. I am ordering the card tonight.
Just explain to me please: which application exactly has to be started for mch operation? Is it included in SDK folders? Could I script it so as to run with windows permanently?
I can be wrong as I’m no audio expert, but I think that to use a multi channel audio feature it will be needed:
1- Hardware support
2- Driver support
3- Application support
The problem it's to me the application itself that needs to support the feature and the wdm driver support in this case to support the Magewell card that do support this feature natively.
As I view the situation but I can be wrong , I think there isn’t a specific application that must me started to initiate multi-channel support , it is supported natively by the Magewell card, and what you do need is an application that do support multi-channel and wdm audio class devices.
The applications that I know that can connect to the wdm audio capture pin are the utility multiaudiocapture that comes as an example in the SDK, the Magewell capture express application that is provided for free to capture and stream video and audio multi-channel ( only mp4 and AAC), virtualDub2,and FFmpeg. There are a large list of magewell compatible software for capture and streaming perhaps some of them will do multi-channel, but I don´t know.
If the application only supports wasapi, direct sound or MME it will connect I think to a specific capture pin in the Magewell and it will be limited to stereo 2 channel , if the application does support wdm and multi channel itself it will probably connect to the magewell wdm pin and you will get multi-channel support.
shekh the virtualdub 2 developer as made the necessary modifications to VDub2 to support wdm audio class devices and Magewell cards in particular perhaps if you contact him he can clarify what is necessary to connect to the magewell wdm capture pin to assure a successful connection for using multi channel.
Perhaps Dante Via does support wdm and can connect directly to the magewell wdm pin?
GraphStudioNext also does support the Magewell Wdm pin connection and perhaps you can use it for your purpose .It's a graphical environment where you can design and test different workflows ( see an example for capture some posts before and a new one
[Attachment 51863 - Click to enlarge]
Last edited by FLP437; 4th Feb 2020 at 20:59. Reason: graphstudioNext inclusion
After FLP437's screenshots I concluded that in my case the card could work even out of the box.
Anyway, I ordered the card.
I will post the results.
FLP437, you are always very analytical and willing to help. Many thanks again.
the card has arrived, I installed it. Works fine.
However the news for the 8 channel output are not good for me.
No matter what method I tried the card will either output front L/R channels or downmix to 2 channels.
[Attachment 52033 - Click to enlarge]
Which is a shame really because the 8 channels ARE THERE! You can even see their waveforms!
In terms of output the Capture pro is recognized as 2 channel device.
I cannot get it to be recognized as 8 channel audio.
I can see that the 8 channels are being succesfully captured but I take as output only the front left/right when using Capture Express or MultiAudioCapture.
Using the DS graph I get a downmixed 2 channel output.
Again, pls see the second screenshot in Dante Controller, where even the JRiver Media Center, which is merely a software engine, can be recognized as 6 channel device, and it works pretty well too.
What really puzzles me is the fact that even using the SDK included 'MultiAudioCapture' example, the multichannel LPCM signals are being captured succesfully but only the front 2 are destined to the output.
I had a look at the multichannel enabling directions but I am not a developer, I am a field-application guy, so I can hardly read these, let alone to understand them.
I'm not sure what might be happening.
Recording works fine with several tools multiaudiocapture, magewell capture express, virtualdub2 , graph studio next, etc
Have in mind that there are two audio capture pins in the card and for multi channel audio only the wdm do works.
Perhaps dante and other applications are connecting to the default pin that doesn't do multichannel.
After a frustrating period I found the solution. This magic driver which, apparently, was quite ironycally already on this site https://www.videohelp.com/software/Multichannel-ASIO-DirectShow-Renderer
[Attachment 52216 - Click to enlarge]
The driver when loaded will present not just 8 but 16 channels on my DANTE controller. All channels are normally seperated, it is a true multichannel streamer and, moreover, it is ASIO.
There still is another serious problem. After some time of operation (15 maybe 20 minutes) the sound gets distorted. It starts with fragmented pops and clicks and soon enough becomes seriously distorted, as if you play in very high levels.
I have to stop the studiographnext, close it down and open it again. After relaunching the sound is perfect again, then after some minutes you start to get pops and clicks > distortion etc...
It is as if you try to reproduce an audio file with very small buffer size and the device cannot handle it.
I am not sure if the problem was overcome or not because there are two conflicting information, this one in the videohelp identifying problems and another one on the blog of the driver developer post by you where everything is apparently solved.
Assuming this is the more up to date info and you are really experiencing problems after some time could it happens that the driver is the solution and the problem simultaneously?
Reading the driver developer blog could it be that you are using settings that the driver can´t cope with ? For instance they do say “ The renderer uses WdlResampler from NAudio library, and as such in sinc-mode it is too slow to cope with 192kHz” .
There are also indications on the same blog of problems with lav filters after 25 min which seems to have some similarity with the indicated problem, others seem to have minimized the problem with the insertion of the filter Asio 4ll in the setting of the filter rather than using it directly ( the external sound card brings a little better sound no cracks, but always a phase shift).
If I were you I will put the situation directly to the driver developer to see if he can identify the problem.
Last edited by FLP437; 5th Mar 2020 at 10:20.
Thanks for your answer!
After some thought research and tests I am looking into this https://github.com/cplussharp/graph-studio-next/issues/291
I think problems in stable reference clocking is the main reason also for the other similar issues (including phase shifts)
Stop and restart the filter in Graphstudionext and you have perfect sound/channel separation/phase, until the point that small repeated clocking errors are being accumulated with all the unwanted results. This point is not specific. Could take 2 minutes of operation, sometimes 10 minutes of operation or more.
Apart from the fact that I am using 8 channels I don't think I am pushing the driver to its edge as I operate @ fixed 48khz.
I will look at it again, try to gather evidence if possible and then I will contact the developer.
Do you agree that clocking could be the problem?
No, it is not fixed.
It is surely much-much better, but definately not fixed.
Phase shifts may occur again. The frequency they appear does not compare with the previous mess. But they'll come at some point again. Rare pops appear too.
Approx 6-7 hours of streaming witho NO phase shifts, just very rare pops, in fact none during the last hour.
[Attachment 52380 - Click to enlarge]
However, there seems to be enormous processing as shown there in sync correction. Cpu usage 7% to 8% (!) and my processor is an i7 7700K..
Yes, I can definatelly listen to the music.
No, this can't be a permanent solution.
It is a sync problem.
I will send my findings to the developer.
Thank you @FLP437 and @stamatiski for all the work you have done here, in particular documenting and sharing the whole process .
I'm also trying to figure out a process for extracting the audio from a HDMI source, with the end goal of getting it to a USB DAC as PCM. I have no interest in the video source at this time.
For example, here is what I envision in more detail:
1. HDMI 2.1 TV - Source device
2. HDMI 2.1 cable from TV's eARC output to a capture card's HDMI input [Which probably needs to be HDMI 2.1 compliant - to be sure]
- eARC is important here because I want to be able to get all raw high-res formats such as Dolby Digital Plus (up to 7.1 channel), Dolby TrueHD (up to 7.1 channel), Dolby Atmos, DTS-HD High Resolution Audio (up to 7.1 channel), DTS-HD Master Audio (up to 7.1 channel), DTS:X etc..
- Would the Magewell Pro capture card work with eARC? (i.e. can the above high-res formats be captured, or do we need to wait for a fully compliant HDMI 2.1 capture card?). Perhaps we make the eARC work on the capture card with some EDID tweaking?
- Will HDCP be a problem? [From what I understand, it can be worked around with the Magewell capture card.. ]
3. Use a software decoder/media player, such as JRiver and configure it to use the capture card's Windows Audio output device
4. Use JRiver to decode (or bitstream) and customise the audio to my needs (e.g. Custom downmixing, room correction etc..)
5. Use JRiver to output PCM to the DAC via USB
Given my current understanding, I'm not sure if this process is fully realisable yet, but I'd appreciate your insights and comments in regards to what I'm trying to achieve. Thanks for your time!
Last edited by Feyire; 21st Nov 2020 at 10:01.
Blackmagic (Decklink) can do it with the Media Express app. QT uncompressed AVI uncompressed (Also AVImjpeg) with 8 (16max) audio channels
Ignore list: hello_hello, tried, TechLord, Snoopy329
Both Magewell and Blackmagic would capture multichannel audio. However the Magewell is more versatile at it does support for customized EDID and can capture in any audio format from 16 to 32 bits and from 44,1khz to 192Khz( I think it goes up to 384k but I´m not sure in this moment). Also in case it will be necessary to capture video even if only to retain afterwards the audio it will provide support to all, uncompressed , lossless or lossy , at any resolutions and color spaces, videos formats the Blackmagic has a much more limited support of formats typical the more usual broadcast formats.
If there is any instability of the source ( unlikely over hdmi , but usually in the analog world) the magewell is a lot more tolerant that the blackmagic also the drivers are very stable which sometimes is not the case with blackmagic.
These cards don’t provide audio decoders themselves so they can only record lpcm uncompressed audio. The magewell card can capture audio only ,no need to capture video if this is not required. Audio capture interface is based on DirectSound, expansion interface based on IKsPropertySet
A small utility included in the SDK Multi Audio capture can capture only audio in exactly the native format of the stream for instance 24 bit / 96000 Khz or whatever the native format is. Support for capturing of IEC60958/IEC61937 audio, including uncompressed audio such as 5.1 channel, 7.1 channel, DTS, THX, SRS and compressed audio such as AAC and MP3
The Magewell presents itself by default to the source as a capable lpcm multi channel audio device. In this context the source will try to decode compressed streams like DTS, Dolby, etc to LPCM which the card will gladly captures.
If the source for some reason is not able to decode the compressed audio signal and provide LPCM multichannel, one possibility is to change the EDID to present the card as a capable Dolby or DTS decoder device. In this context the source will provide the raw dolby stream for instance and we need to capture a bit accurate audio stream and decode it externally to lpcm.
If the audio sources don’t provide lpcm ( with some is possible also to change audio settings in the menu to force lpcm output) we have to capture the bit accurate raw audio stream and decode it afterwards if possible.
In the previous posts has been detailed the process to capture the Dolby stream with virtualdub and decode it afterwards with ffmpeg . FFmpeg will decode most of compressed multi channels formats but I think not the most recent and esoteric ones.
Other possibility that I have not explored but should work for Dolby is the AC-3 ACM Codec, is an AC3 ACM decoder. Install this and you can theoretical convert/decompress AC3 audio directly to WAV/MP3 using Virtualdub, and other applications.
All of these cards are HDCP sensitive, they will not record by default from an HDCP protected source.
It will be necessary to interpose a device in between the source and the card to strip HDCP. Two inexpensive devices (about 20 €) are usually used for this an HDMI splitter (but only some of them will work) or a hdmi to hdmi with audio extractor (usually 2 channels stereo). If you find the correct ones they will work flawless.
Typical the 2k capture cards are hdmi 1.4a or b and the 4k capture cards are hdmi 2.0 as there is no need for more. probably a 8k capture card will be for sure hdmi 2.1
I think but I'm not sure that the major problem is if you go up and exceed the hdmi bandwidth the card can support or if you try to use some feature specifically to a higher hdmi version, otherwise even if source and card have different hdmi versions , probably it will work.
I have a 4k HD Blu-ray recorder, hdmi 2.0 compliant and I have no problem capturing from it with the magewell HDMI pro supporting only 1.4a capturing up to 2k the maximum the card can capture and with an hdmi splitter theoretical supporting only 1.3. I cannot use it to capture 4k that’s for sure, for that I will need a 4k capture card hdmi 2.0. Also older cables are compatible with new hdmi 2.1 devices thought they will be limited to the bandwidth they can carry for instance 18.0 Gb/s and the functionalities possibly to provide up to this bandwidth value and anything that will need full 48 Gbit/s will not work or will be downgraded to be supported by the 18 gbits actual cable bandwidth .Has you are interested in audio only you can always downgrade the video to assure these limits are not reached.
About support to eARC I have no idea and I cannot test as I have not one of this fancy outputs but if you can force the source to provide lpcm instead, probably you will have no problem if not you will be in trouble for sure namely with newer compressed formats as DTS:X and Dolby Atmos..
You can also use an external device a HDMI audio De-Embedders not sure what the correct one for you but for instance
The problem could be you will not find one that covers all the decoders you pretend to use..
Last edited by FLP437; 22nd Nov 2020 at 15:27.
This is interesting to know, thank you for sharing.
Side note: Can the magewell card buffer the incoming data? For example, if you wanted to introduce an arbitrary delay on the HDMI pass-through output?
Again, thanks for your help @FLP437
• I don’t have a Magewell usb dongle so I can´t test it. However consulting the specs for the Magewell usb devices it seems they only support stereo recording they don’t seem to support multichannel and they do seem to also be limited to 24 bit / 96 Khz. To get multichannel audio and 32 bit / 192 khz audio you do have to use the Magewell Pro Family (PCIe Gen2 ) cards.They do have a different main chip on board.
USB Capture HDMI Gen 2 specs
HDMI Specific Features
• 165MHz HDMI receiver
• Adaptive HDMI equalizer support for cables lengths up to 30M
• Support for customized EDID
• Support for extraction of AVI/Audio/SPD/MS/VS/ACP/ISRC1/ISRC2/Gamut InfoFrames
• Full colorimetry support
• Support for 8/10/12-bit color depth
• Support for RGB 4:4:4，YCbCr 4:4:4，YCbCr 4:2:2 color sampling
• Support for 2 channel IEC60958 audio streams
• Support for extraction of audio formation information & channel status data
• Support for extraction of video timing information
• Support for extraction of 3D format information
• Support for Side-by-Side Half, Top-and-Bottom, Frame Packing 3D mode.
Thanks, that explains well