Lossy formats

May 27, 2016
 by Paul McGowan

I am often asked about the difference between lossy and lossless and what it actually means. I am also asked why our DAC natively only supports one and not the other.

Lossy means to lose something: in this case, musical information. Lossless, means the opposite: all the musical information is preserved.

Why would we want one over the other?

Both terms refer to compression, the practice of squeezing more out of less. In this case, more data in less space. There was a time when memory and bandwidth were more precious than they are today. And back then whatever designers could do to squeeze more data into a smaller space meant more songs could be placed on hard drives, and hard pressed networks could stream music to more people without clogging up.

More, more, more.

There's only so much you can do to squeeze more into a smaller space if you're unwilling to lose information. Lossless file types, like FLAC and ALAC, squeeze about twice as much data into the same size container as their uncompressed versions. To get more data into a smaller space you need to start giving up some of your data. Lossy.

Lossy files can range from not losing much to losing a hell of a lot, and everywhere in between. The most famous of the lossy files is MP3. What's interesting about MP3 is its variability. MP3 can range from the very compressed to the not so compressed, depending on the intent of the person compressing the data. I've heard MP3 files that squeezed double the amount of data into half as much space as a lossless file and they weren't half bad. Not half bad at all. Listenable, especially if you're not in the critical listening mode.

We'll look at some other file types tomorrow.

Subscribe to Paul's Posts

12 comments on “Lossy formats”

  1. In this case lossless is a most misleading terminus if you haven't been informed about its definition in digital audio. I guess that every digitization is lossy and prone to distortions. Only if the clock and its frequency variations are dynamically identical for both the ADC and the DAC distrtions could be avoided. The process is lossy because it only takes a limited number of samples from the analog signal of the microphone. Thus this process is prone to not detecting sharp transient peaks! The process would be lossless if every sharp transients (and not the limited frequency range of red book cd) that characterize the music are captured! 😉

  2. This subject of course continues to be the source of much controversy and discussion, with, in my experience, neither side bending in their beliefs. Even among the lossless crowd there seems to be minimal agreement, with those preferring DSD trying to convince those that prefer PCM of each of the merits of each (but generally there is not much convincing going on, rather it is a series of pronouncements designed to belittle the other's beliefs for the benefit of their own - a lot like the current political environment). At the recent Axpona show during one demo there were guys being quite argumentative at one another over whether the FLAC or WAV files sounded better or not. Not quite ready to come to blows over it but certainly fanatical in their positions.

    For this listener, and I have said this before, I can personally find no sonic difference at all no matter how much I try between the lossless and lossy formats with the music that I listen to (generally pop, rock, folk and other music from the 70s to current day). In fact, I firmly believe that a straight 320kb mp3 is more than "good enough". I have tried, ad nausea, to detect differences in the songs from FLAC or Redbook CD or mp3 and there just are none in my opinion. So, I live happily with the 320kb mp3 for everything, just ripping my discs to that format and not giving it a second thought. The provenance of most of the music I listen to is such that the mp3 file already exceeds the dynamic range of the source material anyway, so I feel I am not missing a thing. With all my expensive equipment you might consider this a waste to play files of this type on equipment designed to reproduce more, but, I'm OK with that. My local audio guy lost a big sale recently when I was going to get some new speakers, planned to spend a few thousand dollars, but then he made a comment such as "why do you want to spend money on something like that just to play those crap files" so as you can imagine he did not get the business and I happily enjoy my Focal Aria speakers I bought elsewhere.

    Of course, I have never heard anything play in Music Room One and might have a different opinion once I did 🙂 maybe one day I will get the chance. BTW, to me dynamic compression is the far greater evil. Grossly dynamically compressed files sound just as awful no matter what format they come in. Even some of the acoustical/unplugged albums by singer-songwriters I have purchased lately are so dynamically compressed the map of the wave file in Roon is almost a solid rectangle across the whole song, no highs or lows, everything loud.

      1. WAV = 1 and 320kps mp3 = 5
        Associated equipment:
        thinksound ON1 headphones and Audioengine D1 DAC
        I tried it a second time with an Oppo HA-2 headphone amp/DAC and Focal Spirit Classic headphones but no difference for me in what I could or could not perceive.

        I believe that much of what people like to listen to in the high resolution arena also depends on what they are after. I think there is a fine line between listening to your equipment and listening to the song.

        I prefer the song. I don't care to hear the shuffle of the feet of the performer, for instance, I want to listen to the overall song, and come away humming and tapping my feet. Too much of high end for me concentrates on the minute things and takes away the joy of the overall.

      2. For me WAV = 1, MP3 = 1, 320 = 4
        Honestly, I'm embarrassed. With Audeze LCD-3 headphones and a Musical Fidelity DAC. I don't think the tracks were especially easy/demonstrative in most cases. The Coldplay one sounded completely horrible. Maybe my high-end frequencies are shot or maybe I need training. Or maybe my DAC makes everything sound wonderful! 😉 I'd like to hear more honest results from this crowd.

        1. I vote that your DAC makes everything sound wonderful!

          I have to be honest the Coldplay one gave me fits too. So much distortion as a part of the sound tough to pull out details. Fun exercise.

      3. What I learned from listening were two things:

        1) Once I know what to look for, I can spot a 128 kbps because it's the smoothest and an uncompressed one because it is the roughest

        2) In two senses, the test is nonsense because...
        a) it assumes that listeners know - are attuned by ear - to the differences between between loss and lossy. If the point is to prove something about marketing for the masses, fair enough. If the point is anything beyond that, it's the equivalent of those studies that use nonaudiophiles to evaluate systems (Consumer Reports) or to prove that systems all sound the same (that study which is quoted all the time by naysayers).

        b) because the volume levels between each set is not the same, which is a confounder.

        1. I should add a third point, which follows from what Paul wrote, namely not all lossless are created equal, but we don't know how the sample recordings were done.

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram