Home > Views > Industry Views > Lossless Audio: Part One

Lossless Audio: Part One

May 18th, 2009
Two decades after Laserdisc, home-theater audio comes full circle on Blu-ray Disc

If you joined the ranks of home-theater aficionados during the glory-days of DVD, you might not be aware that while DVD improved dramatically over the image quality of previous consumer video formats like laserdisc and VHS, in some ways it took a step backwards with audio quality compared to what laserdisc had offered.  “Hugh???” If that describes your tone of disbelief, I’m sure you’re not alone; most folks trustingly assume that each step “forward” in technology is just that—and objective and quantifiable step forward. And in some ways DVD audio did step forward: it brought discrete multichannel surround sound to the masses which had the advantage of full-frequency and independent surround channels in addition to pin-point directional steering: two things that Pro-Logic-processed 2-channel matrixed recordings couldn’t claim to offer. And yes that’s a step forward. But what you may know is that in order fit 5.1 channels into the space available on DVD, a compromise had to be made that, channel-for-channel, caused the audio fidelity of DVD to fall below the level enjoyed by the CD-quality PCM stereo tracks that had been available on laserdisc for years.

How was the sound of DVD compromised?
I’m sure you’ve heard the term “compression”, and it describes how large data-files are scaled down to a smaller size for storage and/or transmission when space/bandwidth are limited. But “compression” comes in two flavors, lossless and lossy, and the difference is important. Lossless compression means you get back 100% of what you put in: that when the shrink/expand process is complete, you end up with a datafile that’s bit-for-bit identical to the original. That’s the way that you’d zip files on your PC to take up less space on your hard-drive or so that they’ll be lighter weight to send in an email to keep it from bouncing back. When you “unzip” the file it’s restored back to its original bit-for-bit structure without any loss whatsoever in the process. Lossless compression doesn’t impose a loss of quality since you maintain bit-for-bit accuracy to the original.
But the compression ratios offered by lossless schemes aren’t very aggressive (typically only 30-50% reduction in file size), and so if you want to shrink those data files to a more drastic degree, like down to one-tenth their original size as is often the case with Dolby Digital audio on DVD, there’s no way to shrink the file down that small and still restore back all of the data you originally put in. And so we instead use a lossy-compression scheme. With lossy-compression, the software doing the data-reduction is challenged with figuring out which data to keep and which data to throw away. Naturally, if you have to throw away data and never get it back again, you’d like to do so with the least amount of theoretical loss of perceived quality. In the case of audio signals, we use what are called psycho-acoustic models of human hearing to guide the software’s analysis of the musical waveform to help it decide what to keep and what to throw away.
Psycho-Acoustic?
Ever notice how when you reach highway speeds in your car, you find it harder to hear what you were able to clearly hear on the radio at the same volume when driving at slower speeds? With all that wind noise overshadowing the music, do you think, for instance, that you’d notice if we dropped off some of those soft piano decays in the radio’s broadcast recording? Maybe the relatively soft back-up vocal vanished entirely once the wind-noise picked up around 60 mph, so you might not notice if we trimmed that vocal track out of the radio broadcast either. You probably wouldn’t notice these changes because, with the added wind noise you, probably wouldn’t have heard those details anyway.
Of course, in your car you have the option to turn up the radio to try to compensate for the highway wind-noise, but now that you’re following this principle of how adding one sound can block out your ability to clearly hear another sound, let’s pretend that instead of wind-noise and radio in a car, you’re listening to a Rock album in your living room. In this song the intro starts off with an acoustic guitar and solo vocal track that are very clearly discernable… you hear the reverberation and decay in the guitar strings and you hear the low-level detail in the singer’s voice.Then the drums kick in and a back-up chorus floods the track and an electric guitar comes in on top of everything else… while the original vocal track and acoustic guitar still continue just as before. Is the detail in the lead vocal’s voice as easy to discern now? How about all the subtle resonance in the acoustic guitar strings that you heard during the intro? When the drums and bass kick in full force, you might not even hear the lyrics clearly enough to catch all the words. You see, just like wind-noise can cover up the details of the music on the radio, one musical sound can mask another. It’s this principle of “masking” in the way we perceive sound that psycho-acoustic compression models use to determine what musical information can get tossed out without you noticing that it’s gone. Or at least not noticing to the point that it bothers you. Or more likely, so that it doesn’t bother most people.
You see, all psycho-acoustic models of human hearing are based on generalizations about how people perceive sound, and no two people perceive sound exactly the same way. And even if we did, it’s difficult to build a perfect model of human hearing because it’s impossible to gauge exactly what people are and aren’t hearing because the listening tests, the questions, and the answers provided, are often subjective and difficult to quantify (even when executed with empirical methods). And though at very mild compression ratios most lossy/psycho-acoustic compression schemes may result in little or no perceptible audible degradation to the majority of listeners, as that compression ratio is increased to minimize bit-space and more and more data is discarded, the discrepancy between the original uncompressed sound and the lossy-compressed result becomes plain even to “I don’t really care” listeners. What may surprise you even more is to learn is that lossy-compression degrades the sound of even old mono-recordings from movies like It’s A Wonderful Life and The Wizard of Oz. You might also be surprised to learn that the difference between lossy and lossless can be appreciated on even “average” quality gear: you do not need to have an expensive AV system to hear and enjoy the improvement with lossless sound.
What does lossy-compression sound like?
Since the nature of what psycho-acoustic compression does to the musical waveform isn’t linear, the effects are also non-linear and may be hard to describe in words for most listeners (even if they aren’t necessarily hard to hear). If you want to hear for yourself, play your favorite MP3 or iTune song (compressed at less than 200 kbps to make this example easy) on your PC over a pair of even mediocre headphones and then switch to the same song on the commercial (uncompressed) CD album played back over the same headphones via the same soundcard on your computer. You’ll instantly realize that a strange and subtle veil has been laid over the sound of the compressed recording… one that seems to cause musical decays to “vanish” as if they’ve been erased rather than fading slowly all the way to silence as they do on the CD original. The compressed recording sounds less 3-dimensional, less open and airy, and things like cymbal crashes that trail off naturally on the CD sound strangely synthetic on the heavily compressed MP3 or iTune. You may notice that in general piano notes, bells and many instruments like strings have lost some of their realism and sound a little more artificial. Vocals that sound richly textured on the CD may sound a bit congested and fused together on the lossy-compressed recording. A visual analogy might be like the difference between a well mastered DVD picture compared to one that’s streaming in real-time via Net-Flix… still watchable, but somehow less satisfying. Now, whether or not this degradation is enough to cause you to want to increase your compression bit-rate to improve audio quality even if it means not being able to store as many songs on your portable player (I only use Apple-Lossless on my iPod because I can’t stand the drop in fidelity even at the higher lossy rates), few music lovers wouldn’t choose to enjoy the full-fidelity sound quality of the original CD recording if it was a simple matter of having or not having the better sound. So switching gears back to our movie-soundtrack, rather than worrying about which bit-rate might cut a tolerable compromise between space and quality, or about how many listeners may or may not be bothered by the lossy-compression, wouldn’t it be great if we could just enjoy the original full-quality audio in our home-theater instead?
That’s exactly what Blu-ray Disc can deliver. Later this week in Part Two we’ll discuss the low-down on how Blu-ray Discs, Blu-ray Players, and surround-sound receivers handle lossless audio to help you make sure that you’re getting the best bang for your buck in your own Home-Theater, regardless of how much money you have to spend.