A Successor to MP3?

In his story You Can’t Be Too Thin for Slate, Paul Boutin describes aacPlus (AAC+SBR) as a potential successor to MP3 for low-bitrate audio. A proprietary variation of standard AAC (without SBR) is used in iPods and iTunes and is somewhat better than MP3 for some applications. AAC is also used by some radio stations and IT Conversations for studio-to-studio links over ISDN lines. Another variation, AAC-LD (for low-delay) is great for audio over POTS lines. The popular presentation at Pop!Tech 2004 by Malcolm Gladwell was streamed from Maine to IT Conversations this way, then re-encoded into MP3 for downloading. aacPlus does indeed sound good at 48kbps and remarkably good at 64kbps. (The 18-month-old reporty cited by Paul wasn’t able to test 64kbps at that time.)

But for the hard-core audio geeks out there, a few clarifications: The basic AAC and MP3 algorithms are in the same family known as perceptual codecs. The “plus” is Spectral Band Replication which is also available with MP3s as MP3Pro. Also, Paul describes aacPlus’ handling of stereo as L+R and L-R signals. This isn’t all that new, since most modern codecs use this scheme and some go even further. (The EBU’s test document doesn’t say which stereo scheme was used for their tests.)

Paul writes that “Webcasters spend most of their money paying for network traffic,” and refers to a $4,000 monthly bill to support 1,000 listeners. I did that math, and he’s in the ballpark. If those listeners were tuned into a 64kbps stream 24 hours a day for the entire month, the cost would indeed be between $5,000 and $8,000 each month and somewhat less at 48kbps. Of course, with BitTorrent the costs drop dramatically, but BT only helps with downloads, not streams. There are, however, P2P technologies for streaming and others are on the way.

But is Paul correct that “Future digital music players will support the format just as surely as they do MP3?” I’m not as optimistic. Sony, for example, is just now coming around to supporting MP3, and Apple has a lot invested in its proprietary version of standard AAC. My sense is that it will take a long time. Don’t hold your breath.

6 thoughts on “A Successor to MP3?

  1. A “proprietary variation of standard AAC” is used in iPods? That’s true for DRMed music, but certainly not about self encoded, non-DRMed AACs.


  2. Hi, Flynn. Yes, I was referring to the proprietary Apple DRM extensions, but The AAC (.m4b) files I produce for IT Conversations use contain no DRM and are, I believe, proprietary. The only way I’ve found to create an AAC file for which the iPod would retain the most-recent playback position (bookmark) was to use the iTunes encoder. This appears to be a non-DRM proprietary extension.


  3. I had the impression the bookmarking extension is triggered by either the HFS filetype or the extension (m4b instead of m4a). I wouldn’t consider this a “proprietary file format”…


  4. AFAIK (based on many hours of experimenting and testing) the only way to make a bookmarkable AAC is to encode it using iTunes, then rename/retype the .m4a to .m4b. I’ve tried what you describe using AAC files from other encoders and never been successful, hence my application of the ‘proprietary’ label to the feature. I’d be glad to be proven wrong, since using iTines for this is a real pain, so please let me know if you find a non-iTunes solution.


  5. You write: “I prefer this [a lowpass filter] to depending on the Nyquist Theorem to accomplish this task.”

    This is sort of twisted around. If you simply downsample, the energy in the higher frequencies doesn’t go away, but its actual frequency is unrepresentable per Nyquist, so what happens in that it gets aliased down to lower frequencies distorting your signal.

    You will say “but when I downsample audio I don’t notice this.” That’s because nearly all audioprocessing software assumes that you couldn’t possibly want frequency aliasing and performs an automatic lowpass filtering at the Nyquist frequency before downsampling.

    So, either way you’ve got a lowpass filter in there somewhere.. downsampling by itself doesn’t get rid of the energy in the higher bands.

    Not to be pedantic or anything šŸ™‚


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s