First baby-step to MPEG-4 DRM

ISMA (Internet Streaming Media Alliance) has completed its MPEG-4 Content Protection Specification, and it’s now available for peer review for experts in network security, content protection and cryptography.

The encryption method chosen for the new specification is based on the National Institute of Standards & Technology’s (NIST) 128-bit AES encryption standard. Importantly, this method is unencumbered by any additional royalty fees and intellectual property concerns. It’s also compatible with established IETF (Internet Engineering Task Force) specifications.

Many people seem to be misinterpreting this announcement as meaning, “Great! Now I can do DRM for MPEG-4!” However, the specification doesn’t specify a specific rights and key management system, and so doesn’t actually enable DRM (much less DRM interoperability) of any sort. Rob Koenen, president of the MPEG-4 Industry Forum, notes:

I see it as another step toward more interoperability in DRM. But that’s a difficult problem to solve. There are many little steps to be taken on the road to more interoperable DRM and agreeing on encryption is only one of them.

Assuming it passes peer review, the specification is expected to be finalized in June.

An MPEG-4 iPod? No, but…

One of the more interesting smart storage devices introduced at CeBIT was the Archos AV340, a 40 GB video player and recorder with a 640×368 screen. It stores up to 80 hours of content — that’s a minimum datarate of around 1 Mbit/sec — for viewing on the device’s 3.8″ screen or a television. It can also be used as a 3 megapixel digital camera with an optional attachment, and will apparently sell for just north of $500.

That’s the good news. The bad news is that I believe that Archos either lying or confused about their MPEG-4 support. The press release specifically talks about the device’s MPEG-4 codec, and says that it supports MP3 audio rather than MP4 audio — that suggests that this device might just be a DivX AVI player/recorder rather than an MPEG-4 player/recorder, which makes it a short-term solution that I couldn’t recommend to anyone.

Apple to introduce Mac-only music service, MPEG-4 AAC support for iPod

The L.A. Times is reporting that Apple will be introducing a Mac-only music service, using iTunes as the client and MPEG-4 AAC (rather than MP3) as the delivery format. A source told the L.A. Times:

This is exactly what the music industry has been waiting for. It’s hip. It’s quick. It’s easy. If people on the internet are actually interested in buying music, not just stealing it, this is the answer.

Yeeesh — clearly an Apple marketeer using the Times to say what they can’t officially say as an Apple marketeer. Happily, just after that vacuous quote comes this useful perspective from the story’s author, Jon Healey:

That ease of use has music executives optimistic that the Apple service will be an effective antidote to surging piracy on the Internet, sources said.

Other legitimate music services have cumbersome technology and pricing plans — motivated in part by the labels’ demands for security — that make them much harder to use than unauthorized online services, such as the Kazaa file-sharing system.

Bingo. The customer is not the enemy, but the customer wants what they want. People wouldn’t steal music if there was a legit, reasonably-priced alternative that provided the same convenience and freedoms. In fact, commercial services should be able to provide far more convenience with features like better searching, affinity-based recommendations, and fast and reliable downloads. But if the Apple service uses DRM to restrict what users can do — and the L.A. Times story suggests that it will — then Apple’s service is going to suffer from the same lethargic uptake as the others.

Apple has apparently signed up four of the five major labels — BMG (Bertelsmann), EMI Group, Universal Music Group (Vivendi Universal) and Warner Music Group (AOL Time Warner) are reportedly on board. Sony Music Entertainment is holding out because — and try not to choke on the irony — they’re concerned that users will be allowed to put purchased music on CDs. One wonders if it might be better for Sony Music if just stopped selling CDs immediately.

Partnering with Apple is a brilliant move on the part of the labels. Not only do Mac users tend to consume more music than the average Wintel user, but it means that they can avoid supporting Mac OS in their own offerings. And with less than 3% of the PC marketshare, the Apple’s customer base is a good place to experiment with business models.

MPEG-4 Systems license released

MPEG LA — one or two “one-stop shops” for MPEG-4 patent licensing (erm, isn’t that a “two-stop shop”, then?) — has released the MPEG-4 Systems license.

The Systems license the last piece of the long-in-coming patent portfolio puzzle. The Systems license covers the MPEG-4 file format, MPEG-4’s interactive features, the MPEG-4 scene description language (BIFS), and the MPEG-4 object descriptor framework.

The object descriptor framework is an extensible model for describing objects and inter-object synchronization. For example, a news broadcast stream may include a “virtual set” object, an “on-air personality” object and a “stock ticker” object.

BIFS (BInary Format for Scene) lets you specify the spatial and temporal composition of objects. It’s a binary standard based on VRML, which nobody wanted the first time around. SMIL enables the same thing, but it works with much more than just MPEG-4, and for this and other reasons I think that BIFS is doomed in the long run.

The file format is obviously the most important part of the Systems license in the short term. I think it was a mistake to group the file format in with a lot of cruft (some of which, granted, is nifty) that the majority of implementations aren’t going to implement within the next few years. My understanding is that MPEG is in the process of reorganizing the Systems specification into several parts, and hopefully this will allow super-inexpensive (free, ideally) licensing of the file format.

Finally, we don’t know what the terms of the System license are yet. Rob Koenen, President of the M4IF (MPEG-4 Industry Forum), had not seen the license as of MPEG LA’s announcement, but said that he expected it to be less controversial than the Visual license given the draft terms that were released last year.

ISMA sets schedule to finalize MPEG-4 DRM spec

This week, ISMA (Internet Streaming Media Alliance) shared their roadmap for the digital rights management technical specification they’ve been working on for MPEG-4. It uses MPEG-4’s IPMP (Intellectual Property Management Protection) standard, and will work for both realtime and progressive (a.k.a. downloaded, a.k.a. shared) streaming content.

They intend to make it available for review at the NAB (National Broadcasters Association) convention in April, and to release the final specification by the end of June. ISMA has formed a content advisory board (members haven’t yet been announced, but it’s likely that they’re courting movie studios, the MPAA and the RIAA) and I expect that ISMA will want the approval of this board before releasing the final specification.

In another bit of very good news, ISMA has announced that they’ll be introducing a certification program. Vendors that are interoperable with the standard will be able to use a common trademark.

Next-gen DVD: There can be only one (standard)

The DVD Forum has announced that they’ll be choosing the video codec for the next-generation, high-definition DVD standard (tentatively called HD-DVD) by March. Here are the four options in the running:

AVC Also known as H.264, this is the next MPEG-4 video codec. It was created by The Joint Video Team (JVT), which is a cooperative effort of VCEG (the ITU’s Video Coding Experts Group) and MPEG (the ISO’s Moving Pictures Experts Group).

MPEG-4 Video This option is the Advanced Simple Profile flavor of the current MPEG-4 video codec. It’s not competitive with AVC or Windows Media Video 9

MPEG-2 with “enhancement layer” The current generation of DVD uses MPEG-2. This enhancement to MPEG-2 adds a complementary bitstream that makes it more suitable for HD-DVD.

Windows Media Video 9 This is Microsoft’s proprietary video codec, which is derived from MPEG-4 Video.

The DVD Forum’s technical working group has successfully used all three codecs to encode 9 GB of high-definition test content onto a two-layer DVD at bitrates as low as 7 Mbits/sec.

An anonymous source told EE Times that Windows Media Video 9 “doesn’t stand a chance politically” even though it’s technically adequate, and that many chip vendors believe that AVC will win out over an enhanced MPEG-2. This means that — for the first time — content developers may be able to use the same video codec to target mobile devices, PCs, and consumer electronics.

"Mastering Compression" class

My friend and streaming expert Ben Waggoner will be holding his next two 5-day Mastering Compression classes on June 30 and August 11.

These classes are held on the Stanford University campus. They’re part of Stanford’s Digital Media Academy program, so you get Stanford Continuing Education credits for taking the class — that means that your employer might even pay for them!

If you can’t attend,

  • Digital Media Academy: Mastering Compression
  • Posted in MPEG-4, QuickTime, RealSystem, Streaming, Windows Media, Wireless
  • Beware first "MPEG-4" DVD player

    Tom’s Hardware Guide (THG) has always confused the DivX codec (once a hack of an old version of the Windows Media Video codec, now an actual MPEG-4 Video codec) with the MPEG-4 format (which supports several codecs, including the MPEG-4 Video codec). The cluelessness continues in their review of the KiSS Technology DP-450.

    THG’s review of the DP-500 notes that it supports the DivX “format”, which they incorrectly equate with MPEG-4. The player does support DivX-style Franken-AVIs that use MPEG-4 Video (Advanced Simple Profile) for video and MP3 for audio, but it doesn’t support MPEG-4. The DP-500 doesn’t even use DivX for their video (which makes the THG review all the more bewildering), but uses REALmagic’s 8500 hardware decoder instead.

    In review, (1) the DP-500 does not support MPEG-4, (2) the DP-500 does not use the DivX codec, and (3) avoid Tom’s Hardware Guide when it comes to anything having to do with MPEG-4.

    Winner of "NerdTV" MPEG-4 encoder showdown

    Ben Waggoner has been helping Robert X. Cringely with an MPEG-4 encoder evaluation for NerdTV. Ben did all of the encoding, and then stepped out of the way to let Robert and his team pick the best.

    Sorenson’s forthcoming Squeeze 3, a $199 MPEG-4 encoder, was unanimously (all viewers determined it did best on all of the samples) chosen as the best.

    QuickTime’s MPEG-4 encoder came in dead last. Clearly, avoid QuickTime’s MPEG-4 encoder for professional work. | Post announcing encoder showdown | Post announcing winner | Sorenson Squeeze 3 for MPEG-4 announcement

    Via Licensing chosen as administrator for MPEG-4 audio patent pool

    The MPEG-4 Audio Licensing Committee (ALC) have selected Via Licensing as the licensing administrator for the MPEG-4 Audio patent pool. This is a good thing, I suppose, but it’s troubling that MPEG-4 technology and content developers will have at least two “one-stop shops” for MPEG-4 patent portfolio licensing — MPEG LA for video and Via Licensing for audio.

    The M4IF (MPEG-4 Industry Forum) asked the ALC to develop licensing terms for three audio profiles: High Quality Audio Profile, Speech Audio Profile and Mobile Audio Internetworking (“MAUI”) Profile. Additionally, the ALC will be working with Via Licensing to develop terms for other audio technologies that fall outside the three standard profiles.

    The Speech Audio Profile will cover CELP (“Code Excited Linear Prediction”, a speech codec), HVXC (“Harmonic Vector Excitation Coding”, a speech codec) and TTSI (“Text-to-Speech Interface”, a standard way to control text-to-speech functionality).

    The High Quality Audio Profile will cover AAC LC, AAC LTP, AAC Scalable, CELP, ER AAC LC, ER AAC LTP, ER AAC Scalable, ER CELP. AAC is an audio codec that works great for music, and CELP is a speech codec. (To deciper the notation tacked on before and after the codec names, see the paragraph after next.)

    The Mobile Audio Internetworking Profile will cover these technologies: ER AAC LC, ER AAC Scalable, ER TwinVQ, ER BSAC, ER AAC LD. AAC (Advanced Audio Coding), TwinVQ and BSAC (Bit Sliced Arithmetic Coding) are audio/music codecs. (To deciper the notation tacked on before and after the codec names, see the next paragraph.)

    The notation tacked on before and after the codec names are “tools”, which is an MPEG-ism for “features”. The LC tool is “low complexity”, which significantly reduces decoding complexity (say, for low-power devices) at the cost of a small quality hit. The LTP tool is “long-term prediction”, which provides significant quality improvements at the cost of being somewhta more complex to decode. The ER tool is “error resiliant”, and allows you to stream over lossy transports (like RTSP). The LD tool is “low delay”, which is handy for live streaming and other less-latency-is-better applications. The Scalability tool effciently encodes multiple audio bitrates into a single audio bitstream.

    The ALC plans to deliver the MPEG-4 Audio Profile license agreements in Q1’03. | Dolby press release