97

I have friends that are very picky when it comes to MP3 bitrate, and will always look for the 320 kbps version of a file. However, I have never noticed any differences, they sound the same to me. I remember reading somewhere, can't remember where, that the human ear is simply incapable of sensing the difference, even if present.

Can anyone shed some light on the 192 vs. 320 kbps issue?

  • 2
    With the modern internet bandwidths and disk sizes, why bother with mp3 at all if you have good ears and value quality? Lossless encoding with a sampling rate at least 44.1 kHz and a bit depth of at least 16 bits doesn't require that much data, and exceeds human hearing capabilities. Any reduction in those specifications, and the quality will get progressively worse, no matter how intelligently you compress the data. – travisbartley Jul 18 '13 at 8:21
  • 12
    You can not discuss the 192 vs 320 kbps issue without also having a discussion about mp3 encoders. That's because there are better and worse encoders and some may provide tracks which show little difference between 192 and 320 whereas other (bad) encoders certainly have noticeable differences betweens 192 and 320 kps. Check archive.arstechnica.com/wankerdesk/1q00/mp3/mp3-1.html for an interesting read on that. – Pieter B Aug 29 '14 at 12:32
66

Here is one surprising result, from an experiment described in Maximum PC's article "Do Higher MP3 Bit Rates Really Pay Off?":

Its conclusion:

[No other] Maximum PC Challenge has ever surprised us as much as this one. It’s downright humiliating, in fact, that in many cases, we were unable to tell the difference between an uncompressed track and one encoded at 160Kb/s, the bit rate most of us considered the absolute minimum acceptable for even portable players.

Some follow-up testing confirmed our suspicions: variable bit rate encoding makes a tremendous difference in the audio quality results, certainly enough to justify—many times over—the slight file size increase. Capping the bit rate at 160Kb/s in MP3 files can be pretty harsh on a track, but allowing the bit rate to wander upwards during more complex passages—as variable bit rate encoding does—and throttle down during quieter sections captures an astonishing amount of complexity while keeping file sizes down to an impressive minimum.

I myself took a similar test and failed as much as I succeeded in identifying which track was which (160 vs 320), a result which is no better than random guessing. I can hear a very slight difference most of the time between LAME-encoded (--alt preset standard*) MP3 files and CD audio, but only on an expensive system with terrific speaks in a quiet room. For earbuds and car listening it doesn't really seem to matter.

The biggest difference seems to be not in 160 vs. 320 but CBR vs. VBR.

* "Current consensus is that settings "--alt-preset standard" are recommended for most cases. This results in a very high quality VBR MP3s, giving you bitrates around 200kbps, depending heavily on the music. Mellow rap can go much lower and loud heavy metal can result higher bitrates. The quality will always remain very high." — cd-rw.org

Addendum 7 years later

Rick Beato has a great video on this topic, which I just discovered on YouTube: Audiophile or Audio-Fooled.

  • 3
    Is the site reputable? Please use reputable sources. I can find sources like that which say exactly the opposite :-) – Sklivvz May 15 '11 at 13:39
  • 27
    @Sklivvz: Are you asking me if I consider it reputable or if I have a second citation giving someone else's opinion that it is? And will I require a citation for that citatation as well, ad infinitum? Look, it is a citation. After reading their testing methodology, I personally find no fault with it. But you may decide that question for yourself by reading the article I cited. – Robusto May 15 '11 at 13:44
  • 5
    Robert Harley claims blind listening tests are flawed, not saying I agree or disagree whatsoever but I was hoping someone would have written a specific response somewhere. – Christopher Galpin May 15 '11 at 23:15
  • 10
    @Christopher - Nice link. When he says "... blind listening tests fundamentally distort the listening process and are worthless in determining the audibility of a certain phenomenon" he's making a huge call. It's tantamount to saying placebo-controlled randomized controlled trials 'fundamentally distort' the drug taking process in my view. Would be nice to find a rejoinder to this. – user2466 May 16 '11 at 1:41
  • 8
    @ChristopherGalpin that editorial by Robert Harley is ridicolous. He says that blind testing is useless because it doesn't give the results he thinks are right... Blinding is essential in testing something like that, it has been shown over and over again. To me that only reads like a childish rant. – nico Apr 8 '15 at 13:44
39

There is a distinct difference, however whether it's perceivable, and how much, depends on many factors.

Unfortunately I can't access the full articles, however here's the abstract of a very relevant article, emphasis mine.

Mp3 compression is commonly used to reduce the size of digital music files but introduces a number of potentially audible artifacts, especially at low bitrates. We investigated whether listeners prefer CD quality to mp3 files at various bitrates (96 kb/s to 320 kb/s), and whether this preference is affected by musical genre. Thirteen trained listeners completed an A/B comparison task judging CD quality and compressed files. Listeners significantly preferred CD quality to mp3 files up to 192 kb/s for all musical genres. In addition, we observed a significant effect of expertise (sound engineers vs. musicians) and musical genres (electric v.s acoustic music).

So, this study finds that the answer is that above 192Kbit/s, further gains depend on the genre and the training of the listener.

Furthermore another study finds that perceived quality depends on whether you are using headphones or not:

The impact of using loudspeaker versus headphone playback on the subjective quality of compressed audio is investigated. It is shown that reverberation and to a lesser extent cross-talk, which both are introduced naturally in loudspeaker playback, can effectively hide coding artifacts.

This other paper describes the differences between different bitrates and different ways of testing. In all cases it shows a very minor difference between 192Kbit/s and 256Kbit/s and basically no difference between 256Kbit/s and 320Kbit/s.

enter image description here

As you can see, above 192Kbit it becomes quite hard to tell the difference.

  • @boe: correct, there's a very big difference, that's why sound engineers don't mix using headphones. The reference is always assumed to be loudspeakers (generally of three to five different kinds: near field, mid field, auratone, hifi, and yamaha ns10) – Sklivvz May 15 '11 at 17:17
  • @boehj Speaker quality matters as much as -- if not more than -- the input. If you listening to a low quality MP3 on iPod-quality (cheap throw-aways) earphones, you wouldn't notice if it switched to a higher quality input because the earphones suck. But on a quality speaker set up (or really high quality headphones), you can tell the difference. – user2514 May 16 '11 at 1:27
  • 1
    @Tyler - Yeh I know. I'm the proud owner of some KRK VXT6s. When I got these I had to replace about 50% of my .mp3 library as these (generally older) encodes were now unlistenable. Great monitors. That said, in this environment I can't tell the difference, in any musical style, between lame with a --preset extreme fast switch and the source CD. And to be honest I could probably dial that down a bit too but disk space is cheap these days. :) – user2466 May 16 '11 at 1:34
7

It really depends on a number of factors. Some of them are:

  • The age of the listener, the older you get, the less high frequency sounds you can hear. Besides age, there are more individual factors.
  • The quality of the music equipment, on a nice home stereo you are much more likely to note a difference than on an Ipod on a busy street.
  • The quality depends on which encoder program was used. The encoder basically decides which sounds to encode in the audio stream and which sounds to omit. Encoder programs have a psycho-acoustic model that tells them which sounds are important to the listener and which ones not so much. The quality of the output depends a lot on the model (and therefore the software that was used to encode the MP3).
  • The kind of music you listen to makes a difference.

Conclusion: it is not only about the bit rates.

These and more points are addressed in this introductory article from Karlheinz Brandenburg (Co-developer of MP3): MP3 and AAC explained (PDF, 143KB).

  • yes, there are many factors - but they are all objective, and they're not that many, as to say it's not practical to identify those with a significant impact. – Mihai Rotaru May 15 '11 at 19:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .