A Deep Dive into HDMI Audio Performance

[Sythrix]

I can't find any information on that. The issue of jitter is created in the receiver (due to so much activity from the video subsystem), not from the player so not sure if they can influence that.

As to your application, if it is for movie sound, I would not worry about impact of jitter. You are stuck with HDMI and that is the most convenient solution to routing and extracting multi-channel sound.

If you listen to CD sources then also route the S/PDIF output to your receiver and play 2-channel content that way.

Thanks for the info.

I am even more curious about it’s purported benefits now, since by your evaluation, there doesn’t seem to actually be an avenue for them.

Probably just marketing, although if anyone has further knowledge, please tell.

 

[amirm]

@amirm , any thoughts on this HDMI to AES-EBU converter?

https://www.prosoundweb.com/channel...f_worlds_first_7-1_hdmi_to_aes_ebu_audio_con/

It's a 7.1 HDMI to AES-EBU converter. Users include all big audio corporations like Dolby, Sony, Microsoft, DTS, Samsung, Christie, Pixar, Disney, GDC, Kinoton, Qube, DataSat, Hoyts, Kodak, CBS, AT&T, Ubisoft, Rockstar North, Galaxy Studios, BBC, Air Studios, Capcom, SingTel, Park Road Post, BluRay Audio Disc.

It costs about US$1700.

These devices violate the HDMI's HDCP license as it doesn't allow encrypted audio from being output "in the clear." That said, chances of someone going after them is negligible.

If you need that functionality, it is fine but I doubt that it gets around fidelity issues of HDMI without measuring it.

 

[jaydear]

Amirm, Great article, I wish I had thought to find it earlier! I have been experiencing poor HDMI sound for ages. Initially I just had a good old Sansui stereo amp with a pair of Tannoys, fed from the headphone output on a small ASUS laptop which gets its content over a Gigabit LAN. This analogue sound was really very good, but my wifey didn't like the look of the old black Sansui nor the Tannoys (too big and wrong colour), and of course it was just stereo, no surround.

I changed to a JBL sound bar + subwoofer combo, the 'phasey' and distorted sound from which just drove me crazy, and in any case it was unreliable and a total waste of money. When it failed the second time (power supply first, then the Bluetooth to the Subbie) I replaced it with an Onkyo 7.1 setup which I'm horrified to say is exhibiting the same annoying sound. The "grit" I'm hearing is a bit like amplifier crossover distortion - but not limited to quiet signals - mixed with broadband intermodulation distortion, and all this fuzz is affected by the nature of the content. Some barely has it, some is almost unlistenable.

It seems the culprit is the ASUS laptop which accesses our videos over the LAN, just using VLC or PotPlayer. The HDMI picture is great, no complaints, but the HDMI audio is very disappointing. If I access the LAN content through our Sony BD player the audio is markedly better, but not as good as it is straight off a BD.

I'm wondering if you have thought of measuring tones from say, an MKV or an MP4 video with perhaps some colour bars or unrelated video through such a Home Theatre PC system which must be pretty common these days. I hadn't given it much thought, but now I realise that my ASUS laptop being used as a HTPC has to convert the audio and video from the mp4 / mkv files on the NAS to HDMI and that like anything else electronic there must be different quality chipsets for that purpose. I think there is a huge potential for very inferior HDMI sound in this common sort of setup whether the source material is originating from a NAS, a local HDD or a USB thumbdrive just plugged into the HTPC. It would be fantastic to get some idea of what HDMI chipsets are high quality and what manufacturers are making Laptops / mini PCs / etc. that can produce the high quality HDMI audio we are seeking.

 

[vert]

I have an AVR incoming (hopefully my last gear acquisition in a long time) with network capacity, I'm curious to see how music will sound over LAN, Wifi or Bluetooth compared to the DAC + stereo amp setup I was using until now for both music and TV, and to the DAC connected to the new AVR. If little difference is heard, I will streamline and put the DAC on sale. Of course one thing the AVR has over the traditional amp is room correction, so I'm not sure how meaningful a comparison will be.

 

[Willem]

To be honest I think the designers of these mass market AVRs hit it just right. They did not waste resources on something that would not be audible. If they had had another few dollars per unit to spend, where should they have spent it? My bet is on a better less anaemic power supply. These are brutally competitive markets where the designer has to get the balance just right. The remaining question is, of course, if better hdmi quality demands more expensive components (a variable cost) or better engineering that just takes more time (a fixed cost).
On a similar note, I wonder what compromises are necessary in mass market stereo amplifiers with digital inputs, like the Yamaha AS 501/701/801. Are their digital circuits similarly compromised, or is it only an issue with complex and noisy AVRs and hdmi?
Of course, all this is different for professional gear, or for more ambitious audio with fewer cost constraints. For me personally, the beauty of digital is that if all is well you do not have to worry about the technical quality of the signal that goes into your amplifier. For that, signal quality that is well above suspicion would be reassuring, and is what I demand. For that reason I fear the day that Google decides that the video Chromecast and its hdmi output is all that we need.

 

[graz_lag]

I have an AVR incoming (hopefully my last gear acquisition in a long time) with network capacity, I'm curious to see how music will sound over LAN, Wifi or Bluetooth compared to the DAC + stereo amp setup I was using until now for both music and TV, and to the DAC connected to the new AVR. If little difference is heard, I will streamline and put the DAC on sale. Of course one thing the AVR has over the traditional amp is room correction, so I'm not sure how meaningful a comparison will be.

If your AVR has the pure direct or straight mode, you can by-pass all the DSP/EQ functions and get straight the 2.0 signal to the amplifier section.
The use of DSD functions such as live/club or entertainment might add some fun instead if you want to DSP your music simulating concerts or such.
IMHO & Experience :
LAN - is very stable and consistent; expect some fractions of seconds between the moment you click play on the player and the sound coming out from the system
WiFi - you may have some drop outs depending the power available of both the network and receiver WiFi device, as well as the available bandwidth; in my systems I cannot hear any difference over the LAN, though
Bluetooth - it is OK up to 320kbps, drop outs occur easily if you walk way a bit from the receiver with your transmitter i.e. smartphone

I do not know what AVR you bought but be prepared to face some minor discomforts with the use of the Android/OS applications for your AVR.
No matter the brand, either one suffer of some issue.
Personally, I have ended up using exclusively BubbleUPnP to manage the flow between the NAS/PC sources and the renderer's in my systems.

 

[vert]

If your AVR has the pure direct or straight mode, you can by-pass all the DSP/EQ functions and get straight the 2.0 signal to the amplifier section.
The use of DSD functions such as live/club or entertainment might add some fun instead if you want to DSP your music simulating concerts or such.
IMHO & Experience :
LAN - is very stable and consistent; expect some fractions of seconds between the moment you click play on the player and the sound coming out from the system
WiFi - you may have some drop outs depending the power available of both the network and receiver WiFi device, as well as the available bandwidth; in my systems I cannot hear any difference over the LAN, though
Bluetooth - it is OK up to 320kbps, drop outs occur easily if you walk way a bit from the receiver with your transmitter i.e. smartphone

I do not know what AVR you bought but be prepared to face some minor discomforts with the use of the Android/OS applications for your AVR.
No matter the brand, either one suffer of some issue.
Personally, I have ended up using exclusively BubbleUPnP to manage the flow between the NAS/PC sources and the renderer's in my systems.

Thanks for the info! It's a Marantz slim model. Their apps are said to have issues so I'm expecting that. It does have a direct mode. I'd like to have Audirvana manage the stream from the PC, once I figure out how that might work.

 

[graz_lag]

Thanks for the info! It's a Marantz slim model. Their apps are said to have issues so I'm expecting that. It does have a direct mode. I'd like to have Audirvana manage the stream from the PC, once I figure out how that might work.

That will be very simple for you to set up :

Audirvana will be working as both media server (the software which manages your music library) and media renderer (your music player).
The Audirvana application for iPhone (A+ Remote) will be your control point by means of which you control the media renderer (your music player).
The Marantz's application will purely and solely be your remote control for the AVR (power on, power off, standby, input selection, and so on, as well as volume control, however adjusting it via Audirvana (music player or control point) will be more practical.

 

[vert]

That will be very simple for you to set up :

Audirvana will be working as both media server (the software which manages your music library) and media renderer (your music player).
The Audirvana application for iPhone (A+ Remote) will be your control point by means of which you control the media renderer (your music player).
The Marantz's application will purely and solely be your remote control for the AVR (power on, power off, standby, input selection, and so on, as well as volume control, however adjusting it via Audirvana (music player or control point) will be more practical.

Great to know! I have the new Windows version of Audirvana, and an Android phone, but maybe the app can control a Windows PC, in which case I can always use an old Apple device from a family member.

 

[graz_lag]

Great to know! I have the new Windows version of Audirvana, and an Android phone, but maybe the app can control a Windows PC, in which case I can always use an old Apple device from a family member.

It looks the A+ Remote application cannot control the current Windows Audirvana Plus release but it will able to do that for an upcoming updated release.
Bubble UPnP application for Android is free, you could always try it and see whether it can hook up with your PC-based Audirvana music player.
If YES, bingo !!! If NOT, you will need to wait for the updated release ...
I am not familiar with the latest Audirvana Plus, which integrates the UpnP compatibility, the old version I used was the 2.2 with no network playing capabilities ... migrated since into the Linux world ...

 

[vert]

Will download that. Whether it works or not, lots of fun experimentation to look forward to. Thanks again, you saved me a lot of time!

 

[amirm]

I'm wondering if you have thought of measuring tones from say, an MKV or an MP4 video with perhaps some colour bars or unrelated video through such a Home Theatre PC system which must be pretty common these days.

If by that you mean files created from ripping blu-rays and such, there are a lot of perils there as far as what audio stream is extracted, whether it is left as is or recompressed, etc. MKV and MP4 are just containers so they don't by themselves explain the nature of the bits/fidelity in them.

 

[jaydear]

No, not rips. It is possible to generate pure tones in e.g. Audacity, SoundForge, etc. which could be used in various ways to measure any distortions, noise, crosstalk, jitter, etc. that may be happening in a LAN to HDMI connected system. I understand the container concept and only casually suggested mkv and mp4 as they are among the most common well-supported containers people would likely be using in their Home Theatre setups. The exact contents and technical parameters of whatever containers were used would be up to you, but the containers and their contents should reflect some typical encoding parameters.

A simple listening experiment I have performed is to listen to a program from Netflix through the Sony BD player and the laptop, switching from one to the other with the Onkyo amp. Both units are connected to the Onkyo through identical 1.5 metre HDMI cables. Though it is not possible to do a quick, clean A-B switch with HDMI, there is an obvious audible difference between the output from the Sony BD player and the ASUS laptop, and that difference is affected by the program content. It seems to me that there is a difference in how the conversion to HDMI is happening in both items. The source program material is identical and from the same LAN so imho there should not be a difference. I think we assume that digital is not corrupted as it passes through a system, but that's not my experience.

 

[DownUnderGazza]

A year ago I signed up to Netflix Premium for the Dolby Vision UHD content on my new 4K UHD DV TV. I decided that I liked the DV colour richness and thought I might have heard a subtle improvement of the UHD soundtrack. Was the 25Mbps video stream really better or not? Despite what I thought I might have heard, was I really only getting 192kps AC-3 audio stream? I had no way of knowing.

Picking up on Amirm's original investigation into HDMI Audio performance, my thoughts are this:
-- can the audio quality of movie and TV streaming services, such as Netflix, be measured?
-- In the light of Amirm's observations of poor HDMI audio measurements above, what in 2019 is the best way to extract a quality audio stream from the TV / streaming service? Optical Toslink? HDMI ARC? 2nd hand $30,000 Mark Levinson DAC?

My current set up is:
-- 4k UHD TV -- Netflix App --> 1080p or 4K content --> Toslink --> Matrix-Digi Mini-i Pro DAC (reported as being 16/48khz)
-- Mac Mini media server --> iTunes / Spotify / Safari + Netflix --> internal upscaling to 24/96 --> USB --> Matrix-Digi Mini-i Pro (24/96)

Addendum
Years ago I ripped all my DVDs with HandBrake (https://handbrake.fr) taking care to retain both the highest quality stereo and original surround sound track available on the original disk. Yes, I spent ages trying different settings to preserve the best quality audio & video I could. Aside from having preserved lossless LPCM audio wherever I could find it, the SD video now looks decidedly VHS-like...

 

[amirm]

Picking up on Amirm's original investigation into HDMI Audio performance, my thoughts are this:
-- can the audio quality of movie and TV streaming services, such as Netflix, be measured?
-- In the light of Amirm's observations of poor HDMI audio measurements above, what in 2019 is the best way to extract a quality audio stream from the TV / streaming service? Optical Toslink? HDMI ARC? 2nd hand $30,000 Mark Levinson DAC?

The situation may be changed now. I have a rather high-end pre-amp/processor ($4,500) coming in so I will have a chance to compare HDMI vs S/PDIF on it.

As to measuring fidelity of Netflix streams, it is not possible because a) they don't have any test streams and b) use lossy compression which has dynamic performance.

In general, I am not too picky about movie soundtracks as I am with music.

 

[DownUnderGazza]

Thanks, I look forward to your findings on this new Pre/Pro.

 

[Blumlein 88]

https://www.audiosciencereview.com/...ntz-avr-7701-dac-measurements.3485/post-84218

Here is a post where I compared HDMI and SPDIF on a Marantz 7701 pre/pro. Not much difference. I think you do need HDMI measures to confirm, but it isn't a given that HDMI is poorly done in good gear. Looking forward to Amir's results too.

I also had the same comparison for an Emotiva UMC200. The HDMI was much, much better than the SPDIF input (whether coax or toslink). I can only surmise that was a poor design issue and nothing intrinsic to the inputs.

 

[DownUnderGazza]

Thanks.
So watch this space then?

 

[JoachimStrobel]

Amir,
Thanks for these measurements and explanations. I like to read about the difference between Spdif and Hdmi in technical terms. What is it that makes them different? How is the signal transported? I noticed that a 15m Hdmi cable will give distorted colors which I do not understand given that it is all digital, whereas color shifts look like an analog problem. So how does Hdmi do what it does.
I guess this important too for the quest of investing in high priced multi channel DACs that work via HDMI if fed externally (like an Oppo 105/205) or keep it midrange due to Hdmi’s flaws.

 

[amirm]

In HDMI, audio is slaved to video. It is embedded in the "unused" time slots for video. What this means is that to receive audio, you must also receive video and then extract the audio. Lighting up the video circuits creates a lot more activity inside AVRs, causing interference with the sensitive audio gear.

As for losing colors and such, there is handshaking that goes on with HDMI at the start of the connection to determine what the display can handle. Any data loss there can cause what you are mentioning.

Overall, HDMI has been a poorly conceived protocol and implementation. Compatibility issues can easily crop up.

With S/PDIF, all it carries is audio and it is something that has been tuned for years now. So unless something really screws it up, it is much easier to get good performance out of it.