Brian's Brain

Speaking of Blu-ray, its ascendancy will inevitably (and unfortunately) bring the issue of audio-over-HDMI implementation inconsistency to widespread consciousness. While HDMI outputs are currently offered on some red laser DVD players, the audio codecs encompassed by the DVD-Video standard are few in number and moderate in bitrate, therefore also supportable by S/PDIF. Since HDMI is a relatively recent addition to the DVD platform, and its presence on DVD players is predominantly restricted to higher-end units that upscale the 480-line source material to 720- and 1080-line resolutions, most consumers continue to employ S/PDIF as their player-to-destination connection. Which is a good thing, because support forums are clogged with posts from harried consumers who can't get audio-over-HDMI working with their DVD players, reflective of the sonic immaturity of initial generations of HDMI transmitters and receivers. That is, if the consumers can get HDMI to work at all...

The audio destination mentioned in the preceding paragraph is usually the A/V receiver, not the display, by virtue of the latter's miniscule speakers and lack of true surround sound reproduction capability (another reason why audio and video historically route over separate tethers, even if HDMI could conceptually comprehend them both). And with DVD-Audio and SACD, of course, S/PDIF wasn't allowed to transport the audio stream; instead, Hollywood paranoia forced the player to natively decode the audio and pass it to the receiver over analog connections, with oft-disastrous speaker management consequences. Audio-over-HDMI is more palatable in the blue laser player generation, for several reasons:

1. Most displays offer only one or two HDMI inputs, thereby encouraging consumers to use their HDMI-inclusive A/V receivers as video switches for multiple content sources.
2. HDMI enables the transport of high end audio formats (lossy- and lossless-compressed, along with uncompressed) supported by both Blu-ray and HD DVD, and its HDCP digital rights management scheme addresses Hollywood copy-protection requirements while not forcing quality-degrading successive D/A and A/D conversion steps as the undesirable solution. In fact, analog outputs are being completely phased out, again to address studios' copying concerns.
3. HDMI is a Hollywood-approved transport for both DVD-Audio and SACD bitstreams.

As you can see from the above links, however, the list of audio formats whose processing needs to be supported in the HDMI source and/or destination is significantly longer in the blue laser era, thereby further complicating scenarios such as:

1. How do the source and destination communicate respective decode capabilities and subsequently, successfully negotiate which of them will handle the audio processing tasks to be performed?
2. If decode is to take place in the blue laser player, how does the A/V receiver communicate user- or autocalibration-configured characteristics of the connected speakers (size, crossover points, spacing) and listening environment ('sweet spot' location) to the player so that it can make appropriate sonic adjustments...and just as important, does the receiver then correctly 'get out of the way' and avoid making undesirably redundant adjustments?
3. And is the player even truly outputting the 24-bit and/or 192 kHz audio (for example) that it's supposed to be delivering to the corresponding HDMI link destination? Or is it decimating sample size and/or rate in a manner that, while (arguably) audibly imperceptible, still violates spec spirit?

Although HDMI theoretically provides mechanisms to handle such scenarios, voluminous and consistent feedback I've received from silicon providers and system implementers alike suggests that vague and incomplete HDMI documentation makes the implementation of this vision difficult if not impossible. There's a reason, I guess, why the HDMI specification is at v1.3b with no end in sight.

In the absence of reliable transducer and environmental information from the A/V receiver, the blue laser player tends to provide duplicate settings. When I had my Toshiba HD-A1 S/PDIF-tethered to my JVC receiver, for example, I redundantly configured my speaker attributes at both ends of the digital audio link, and I frankly was never completely sure which (if not both) piece(s) of gear was/were doing the bulk of audio processing. And my poor friend Terry, who inherited the HD-A1 from me a few months ago, has spent many frustrated hours (including face-to-face, telephone and email consultation with yours truly) getting audio-from-HD DVD player working, along with bigger-picture achieving reliable HDCP handshakes between the player and Samsung plasma display (which he also inherited from me).

I wish I could tell you that Audio Precision's APx585-with-HDMI analyzer will solve all of your HDMI protocol problems, but AP's spokespersons whom I met with this week at NAB say that's too big a challenge for the company's modest headcount to tackle (at least for the moment). What I can tell you with a reasonable degree of certainty is that the APx585 will greatly assist your audio-over-HDMI debugging tasks. I followed up my Monday lunch meeting with AP by dropping by the company's booth on Wednesday, where I was treated to several very impressive demos of the product's (along with companion PC, which actually does much of the signal generation and analysis heavy lifting) capabilities.

When I pointed out that, based both on overall industry observation and personal experience, the HDMI ecosystem could have really used this product several years ago, I was bemused by AP's response. Pointing out that AP was a chip implementer, not an IC designer, the company's representatives agreed with me but deflected blame for the delay, pointing out that only now were they able to obtain bug-free audio-over-HDMI chipsets that enabled them to implement their longtime vision. Best of wishes with the APx585, Audio Precision, it's a much-needed offering. And as your personnel's bandwidth enables, please consider adding support for video, HDCP and other protocol signal generation and analysis...even though, I know, 'Audio' is notably featured in your company name.