Lossy Audio Is Acceptable When You Understand Its Limits
With the evolution of video download services, discussions of video compression have become wildly popular. Maybe it’s me, but it seems like audio is being disproportionately ignored.
DYNAMIC RANGE COMPRESSION
There are two different kinds of audio compression: dynamic range compression (DRC) and data compression. If you aren’t sure what dynamic range is, it’s the difference between the noise floor in the
Audio compression artifacts were most noticeable on human voices. room and the loudest sound that the sound system can produce. Typical dynamic range in a top-drawer screening room is 100dB. DRC does just what the name implies; it makes the dynamic range smaller. Peaks in program material are reduced and dips are boosted to make it sound more “even” but less “dynamic.” Only Dolby currently builds DRC capability into its codecs. Some products offer after-market DRC (often called “late night mode”) for DTS and PCM, but it’s provided by each manufacturer individually.
The effects of DRC can be subtle (think talky drama) or blatant (loud action movie), but your clients will be able to hear it. DRC robs a system of its most-appealing attributes: bass and dynamic slam. I recently auditioned an AVR that automatically applied DRC to the Dolby TrueHD track on Iron Man. It made my high-quality speakers and subwoofers sound like an anemic Home Theater in a Box. DRC is really intended for listening late at night, reducing the dynamics of a soundtrack to avoid scaring the kids or grandma, or preventing dynamic digital soundtracks from overloading small micro-systems. Turn DRC off (not automatic, but off) if you want pure sound performance in a screening room.
The idea behind data compression is relatively simple. Let’s say that the original master for a soundtrack requires 100 units of memory to store. Unfortunately, the release format for that soundtrack only has 25 units of memory available. Audio data compression takes that 100 units and shrinks it down to 25.
There are two main categories of data compression: lossy and lossless. With lossy compression, the encoder analyzes that 100 units of memory based on advanced research in the spectral and spatial aspects of human hearing and discards the least significant 75 units in an irrevocable way. Lossless compression analyzes the 100 units and figures out a more efficient way to pack them without
losing any data. However, it typically only reduces the file size by 50 percent. In a few rare instances, studios choose not to use lossless (here’s looking at you, Warner) because the bitbucket might overflow.
The good news is that lossless compression produces no audible artifacts. The track you hear in your screening room is identical to the track that was encoded. Your clients will think lossless tracks sound great. A high amount of lossy compression, though, can be fairly audible. This is especially true if your clients have become accustomed to lossless audio on Blu-ray. I recently spent some time comparing several soundtracks on Blu-ray (lossless DTSHD Master Audio) and Vudu (lossy Dolby Digital). I’m not knocking Vudu, but the lossless tracks blew away the lossy ones.
So where were the compression artifacts most noticeable? Loud explosions? Popping gunshots? Pounding scores? No. It was voices. Dialog through the center channel was the most problematic, closely followed by what were supposed to be airy strings in the score. The top end of both sounded alternately chalky and swishy with the heavy lossy compression. Voices are the most easily recognizable element in a mix for most people, because we hear voices all day long, every day.
RUBBER, MEET ROAD
The vast majority of people will be just fine listening to lossy compressed audio as long as they know its limitations. And that’s exactly how I would approach the topic with your clients. Explain to them that nothing’s wrong, but that they just hear the difference between lossy and lossless sound. If the compression ratio isn’t severe, lossy tracks can sound quite good; Dolby Digital at 448 kbps is awesome for the bit rate. However, the severe file size constraints imposed by release formats like downloads, etc., tempt program providers to use a much lower bit rate, or even down mix a 5.1 track to 2.0—taking you all the way back to 1987 when ProLogic was king. If enough people complain, we’ll see high bit rate lossy or lossless audio pop up in more places.
Chase Walton contributed to this column.
Anthony Grimani (agrimani@ pmiltd.com) is president of Performance Media Industries, a California-based acoustical engineering firm specializing in home theater design and calibration.