Can You Really Tell the Difference Between Music at Different Bit Rates?

(Note: this article is written for Mac users. If you have Windows tools to recommend, please mention them in the comments.)

The bit rate debate regarding compressed music is one that will be around for a long time. Some people think that any compression of music files is anathema. Take Neil Young. He complained about the poor quality of digital music files, while greatly misunderstanding much of what is involved in compression. He claimed that only “5 percent of the data present in the original recording” is present in MP3 files, without specifying the bit rate used or the original sources, and without understanding that compression is more than just lopping off bits of the music. (Andy Doe, writing on the Naxos Blog last year, published an article, All About Bitrates, explaining how compression works. You should read this to understand some points that most people overlook.)

When you start ripping music, and decide what bit rate to use, you have several options. You could go for lossless, which compresses music around 40-60%. One advantage to this is that you can then re-convert the lossless files to a lower bit rate if you want, keeping the originals as archival copies. But lossless files take up much more space. While this isn’t an issue on computers – hard drives are huge these days – it is for portable devices like iPods or iPhones.

If you don’t use a lossless format, you have to decide which format to use (AAC or MP3), and what bit rate. For a long time, Apple sold music at 128 kbps at the iTunes Store. It is now 256k, which is roughly what Amazon uses in their MP3 store (their music is in VBR, or variable bit rate, so it is not exactly 256k). This is an excellent compromise between space and quality. But you might want to go even lower. What’s important is to find the point at which you cannot hear the difference between an original file and a compressed file, and stay above that bit rate.

To do this, you need to perform what is called blind ABX testing. You are presented with music and don’t know which bit rate you are hearing, and you must choose whether you think it is compressed or not. While this test takes a bit of time – you need to rip tracks at different bit rates, then test yourself, one pair of tracks at a time – the results can be interesting.

To start with, find several songs or tracks that you know very well. It’s best to use familiar music, because you will be able to hear more of the differences (if any) because of your familiarity with the melodies, arrangements, etc. I’d recommend not ripping full albums for this test, but rather individual songs or tracks from different albums.

Rip these tracks from CD in lossless format. In iTunes, go to Preferences > General, then click on Import Settings. Choose Lossless Encoder from the Import Using menu.



Next, add the tracks you have ripped in lossless format and to a playlist. Select them all and press Command-I, then enter an album name, such as Lossless Tracks. You’ll want this later to be able to find them.

Back in iTunes’ Import Settings preferences, change to the AAC Encoder, and set the bit rate at 64 kbps. Yes, I did say 64 kbps; you want something that is low enough so that you are guaranteed to hear a difference. As above, create a playlist, then tag all these tracks in an album 64 kbps Tracks.

Go to the Mac App Store and download the free ABXTester. This application lets you choose two tracks, then listen five times to a random selection of those tracks, and choose the one you think is better.



Click on Select A. Navigate to your Lossless Tracks album, and select a track. Click on Select B, navigate to your 64 kbps album and select a track. You know that the first track is lossless, and that the second severely compressed.

The next step is to try five tests, listening to tracks selected at random, and choose whether you think each track is A or B; in other words, which is the better sounding track. For example, when you listen to the first track at X1, if you think it’s the lossless track, click on A; if not, click on B. At the end, click on Check answer, and see how well you did.

I suggested starting with 64 kbps tracks so you can hear a difference. The next step is to find at which point you can no longer hear that difference; at which point your results are no better than random (2 or 3 correct, or a score of 40 or 60%). Rip the same tracks at several other bit rates: I suggest you use 128, 256 and 320. If you do too many, the test will take too long. Label each group of tracks. You can now either go to the next level, 128 kbps, for track B, and go through the tests, or you can start from the other end, at 320 kbps, and work your way down. No matter what, I think you will be surprised.

A couple of notes. First, make sure you do this test on the stereo equipment or headphones you use to listen to music. (If you’re curious, these are the headphones I use.) You might find that, if you try this out on a friend’s Really Good Stereo that you might hear a difference. If you don’t plan to buy the same Really Good Stereo, don’t bother testing on it. If not, you won’t hear a difference; I guarantee it.

Of course, this begs the question: if you don’t have good stereo equipment or headphones, is it worth using a higher bit rate? If you’re thinking of upgrading your stereo or headphones, you might want to plan ahead and do this test on better equipment to see if it’s worth ripping your music now at a bit rate higher than where you can currently hear a difference. And if you have plenty of disk space, you might want to rip your music in lossless format for archival purposes, then convert it to a lower bit rate for use with your portable devices. With iTunes, you can rip in Apple Lossless format, then have iTunes convert your music files during the sync process to 128, 192 or 256 kbps.

Also, if you want to compare high-resolution files, make sure you open Audio MIDI Setup (in the /Applications/Utilities folder), and set the sample rate to the highest possible setting. Otherwise, you won’t hear the full resolution of these files.

I’d be interested to hear your results. Feel free to post them in the comments. If you can really hear a difference between a lossless file and a high-bit-rate compressed file, please also post what kind of stereo equipment or headphones you are using.

68 thoughts on “Can You Really Tell the Difference Between Music at Different Bit Rates?

  1. I’ve done a double blind test with 256AAC vs a lossless file, using headphones and a DAC vastly better than what 95% of people use…. and I had nothing. I didn’t even try to take the test… the files were completely indistinguishable to these 43 year old ears. I’d bet $20 that Neil Young couldn’t tell the difference either.

    I am very, very, very skeptical when anyone claims the difference is “night and day”. The potential for wishful thinking is so high that I simply can’t take you seriously unless you produce some objective evidence.

    There’s no such thing as “the perfect recorded music experience”, because recorded music is inherently a compromise. Enjoy your indistinguishably good compressed audio and spend your money on good speakers and more music!

  2. I have tried stuff like this in the past with my own collection and can’t tell the difference unless the bitrate is well below 96. I haven’t bought a CD for years but have composer friends who will only buy CDs because they feel “digital” music is inferior, not realizing that CDs are digital. Sigh.

  3. Interesting. Sorry but I can tell a difference using lossless files against 256 aac. The lossless files sound crisp and clean and FAT. I am using Klipsch One headphones. I also have a pair of Mackie MR8 speakers I used as a hobby to play techno using my macbook and Traktor. If I buy a new track as a WAV from Beatport and convert to 256 aac there is a big difference in what I can hear. (i Like FAT bass!)
    Storage is cheap now. I have a Drobo with 4x2TB drives in I use this for my iTunes library and all my music is in Apple lossless. I figured that if I was going to make the time to rip my music properly then lossless is a great way to preserve the CD giving me flexibility. I also use iTunes Match for my family’s iOS devices which works great for them.
    I have also DJ’d out in bars and on a few big club systems using Traktor and I am so glad I used lossless files as they sounded fantastic.
    I have the latest Boards of Canada album in 24bit this sounds even better. I hope this is the future.

    • “Sorry but I can tell a difference using lossless files against 256 aac.”

      Are you testing yourself blind, or do you know whether you are listening to lossless or 256aac? If you’re not testing blind, then sorry, but I’m not convinced (of course, you are under no obligation to convince me of anything). Since you left out this important detail of your testing, I assume you are not testing blind.

      • I did some blind tests with various types of music using my HiFiMan He-500. Most of the time I could hear the difference between 128 and lossless. In some blind tests I could even pick the difference between lossless and 320 but that depends a lot on the music and I have to go over the same portion of song multiple times to make sure that I can really hear that minor difference. I did not test any bitrate between 128 and 320. But I can confidently say that 320 and lossless sound pretty much identical to me for almost all songs and there is little advantage for me to keep those flac albums on my hard disk.

      • I’ve tested myself with 320k MP3 vs lossless and picked the lossless song 5 out of 5. You need good equipment, but the difference can be discerned. MP3 also decreases stereo separation, which can be obvious in certain tracks.

    • I’m with Darren on this. I don’t believe one bit of what you claim. Bass is the last thing that’s affected by lower bit rates. In fact, it probably isn’t even affected at all.

      • Actually, if you compare the waveforms, like I did recently with a 128k MP3, the bass (20 Hz) was boosted by 6 dB, which id significant.

  4. Using the stock headphones that came with my HTC One plugged directly into my Macbook Air, I couldn’t tell the difference between Apple Lossless and 128 Kbps using ABXTester.

    I’m looking forward to trying the test with Etymotic ER4Ps, Grado GS1000s, and Mackie HR824 speakers. I’ll post an update after I’ve tried them.

    Thanks for the very helpful post!
    Justin

  5. May I suggest to those of you that try on the proposed abx blind test comparing Cd lossless AAC256 and any lower quality codec : please use well recorded tracks with lots of harmonics (Grand Pianos, violin, etc.) and female voices (in natural or studio environment). This is where you might have a better chance discerning the difference between higher quality uncompressed music (cd, dsd, lossless, aiff) and lower quality codecs (mp3, aac and the rest). Please use blind testing (abx or similar) and tell us your results!

  6. I can clearly hear the difference in my Denon MD 39. The alac files I ripped from my CDs sound wonderful, they deliver a crystal clear and profound sound. On the other hand aac bought from iTunes or mp3s at 320 sound very sharp and shallow!

    • Do ABX test as described in the article.The point is NOT to know which source you are listening thus eliminating the placebo effect.

  7. I recently ripped all my CDs (not a lot, under 3,000 songs/pieces) in 320 kbps variable rate AAC. Then I thought whether I should’ve done it in lossless format instead. So I read to learn, and stumbled upon this great discussion thread. Upon suggestion I picked an album I know so well and have listened to most often (because of countless listening in the wee hours of the morning over the past decade when the wife and kids are asleep), J.S.Bach’s Goldberg Variations played by Konstantin Lifschitz. This is nothing but a grand piano. There are some dynamic passages, e.g., track 25 “Variatio 24” spanning soft and loud measures. I used recently bought Bose QC25 headphones. At 320 kbps variable rate AAC it sounded indistinguishable from the audio CD. So, with peace of mind, I’m keeping it at 320 kbps variable rate AAC. Just thought that someone else might appreciate this testimonial. Now I think that with all their money Apple were able to hire the best physics graduates and sound engineers, and build the best sound labs when they were designing the AAC codec. But I just had to be sure. 8D

  8. Uh-oh, upon further reading I’d like to correct myself. Apple wasn’t the one who defined the AAC codec. Wikipedia says, “AAC was developed with the cooperation and contributions of companies including AT&T Bell Laboratories, Fraunhofer IIS, Dolby Laboratories, Sony Corporation and Nokia.” Something new to learn everyday. 8D

  9. Oh, I just read that too. I had a nagging suspicion right after I posted my first comment today. Thank you for pointing it out clearly.

  10. Just for some context. I have very finely tuned headphones – using a free system-wide parametric EQ (Equalizer APO) and a frequency sweep generator (http://onlinetonegenerator.com/frequency-sweep-generator.html), I tweaked it so that the the entire frequency range (from 20Hz right up to 20,000Hz) sounds equally “loud” without any dips or spikes. This took several days of careful tweaking, but it is completely worth the results. Properly EQ’d headphones will save you oodles of money and get you excellent sound quality – just make sure that if you get an over-ear pair, they have 40mm drivers. That will ensure you will get very robust sub bass down to about 19Hz after a proper EQ.

    One of the things I learned while fine tuning my headphones is that the highest possible frequency I can hear before the tone disappears into noise is around 19,100Hz. So relatively healthy ears, but it’s important because it means that I know any lossy codec that I decide to use doesn’t need to retain any audio information above that frequency, while still sounding subjectively the same as the lossless source.

    If you’re like me and you can’t hear anything above around 19,100Hz, but you want to be absolutely sure that you’re not compromising on any of that high frequency information that you CAN hear, 192kbps AAC (specifically the iTunes encoder) is completely indistinguishable from the source. I mean zero artifacting and zero loss of subjective sound quality. Peaking at the spectral frequency display in Adobe Audition, 192kbps AAC consistently retains frequencies up to 19,350Hz, and pushes up to around 19,500Hz when needed, so there’s significant headroom there for people with even better hearing than myself. Before I switched to AAC I was using the latest LAME MP3 codec at 224kbps, and the lower bitrate AAC file not only easily matches it, but also compresses the waveform much more elegantly and cleanly than the MP3. The MP3 codec seems to have a much more aggressive frequency-removal algorithm, that still for the most part ends up sounding transparent, but takes up more space – meaning that AAC is objectively superior in both sound quality and efficiency (higher quality at lower bitrates).

    To really test the power of AAC at 192kbps in 2017, I would recommend using a track like “On The Dunes” by Donald Fagen, which has very clean production and lots of intricate, ultra high frequency sounds like bells, chimes, ride cymbals and the like. In my experience 90% of music rarely sounds this crisp and clear as does this particular track, so it’s a really great way to ensure peace of mind once you realise your chosen bitrate can do this track justice. In my case, it sounds every bit as detailed and clear as the original lossless version. 192kbps AAC, for all intents and purposes, is CD quality – and in my opinion makes efficient lossless formats like FLAC and ALAC – and even AAC at 256 and 320kbps – obsolete for playback through any sound system.

    TL;DR – For even the most discerning ears, 192kbps AAC (iTunes encoder) gets you CD quality with zero perceptible loss in audio quality; is easily better than equivalent MP3 bit rates; and makes FLAC and ALAC pointless unless you want to keep lossless copies around for archival, or for more lossy conversions in the future.

    • Thanks for the link and performing the tests. How exactly did you use the link to tune your headphones? Did you do it by ear or with electronics (microphone)? One post claimed that channel separation might suffer. Have you tested if that is the case?

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.