It’s got to be one of the most talked about topics in the world of audiophilia: can you hear the difference between MP3 and WAV? We wanted to definitively know as well, so we conconcted a little test to find out, and many of you have participated, listening to the differences — if there are any. And well, turns out not many of you could hear any difference at all. And that’s no surprise.
How does MP3 compression work?
MP3 compression is based on a perceptual model of human hearing acknowledging that the human ear doesn’t equally perceive all frequencies. This leads to the possibility of removing some sounds that are inaudible, thereby saving space without noticeably degrading the perceived quality. The compression process can be broken down into a few steps:
Analysis: The MP3 algorithm first analyzes the audio file to determine which parts of the sound are important and which are less audible. Audio signals are split into different frequency bands because human auditory perception is more sensitive to certain frequency ranges than others.
Polyphase Quadrature Filter: This splits the audio into 32 frequency bands. This step is crucial as it models how humans discern the resolution of sounds at different frequencies.
Modified Discrete Cosine Transform (MDCT): This transform helps in converting the sound from the time domain into the frequency domain. It’s more efficient than previous transforms used in older audio compression techniques because it minimizes the block boundary problems of the traditional Fourier transform.
Psychoacoustic Model: This model is used to determine sound thresholds that can be heard. By understanding which sounds are perceptible, the MP3 algorithm can discard the inaudible frequencies, a process known as auditory masking. In simple terms, loud sounds in certain frequencies can mask quieter sounds in nearby frequencies.
Quantization and Coding: The remaining frequencies are then quantized, which reduces the precision of the sound representation and thus saves data. Finally, Huffman coding — a form of lossless data compression — is applied to further reduce the size.
Encoding: At this stage, all the manipulated data is encoded into an MP3 frame which contains a header and the compressed audio information.
The degree of compression can be adjusted. Higher compression results in smaller files but can affect the audio quality, introducing artifacts like pre-echoes or a loss of clarity in high frequencies. Lower compression ensures better sound quality but results in larger files.
The test
As briefly mentioned above, we concocted a little test to see if our listeners could perceive a difference between the original 44.1kHz 16bit lossless WAV and its MP3 counterpart. The test was divided into five levels corresponding to five compression settings in the MP3 decoder: Lowest, Low, Medium, High, and Highest Quality. At each level, we provided five samples from five different recordings we’ve made, in different genres and different instrumentations. Each track again has two samples, A and B, and it was your guess which one was the MP3-encoded version. So how did you score?
The results
Let’s start at Level 1 (lowest quality compression settings), and work our way up. About 55% of respondents couldn’t hear a difference between the MP3 file and the WAV file. Within this test, the groovy jazz sample with piano, drums and string quintet, apparently was the easiest one to compare, with 62% of you guessing correctly. The classical nonet recording turned out to be the most difficult, with only 52% of you picking out the MP3-encoded version.
At Level 2 (low quality compression), about 57% of you answered correctly, with again the jazz sample being the easiest (71%) and classical nonet the most difficult (43%).
At Level 3 (medium quality compression), the percentage of correct guesses which one was the MP3 drops significantly — only 36% of you guessed correctly. Here, the orchestral and the piano solo samples were the most difficult ones to hear the difference in.
At Level 4 (high quality compression), again only about 37% of you found the MP3 version within the A and B samples. The same samples again proved the most difficult ones.
At the final level, Level 5 (highest quality compression), it shouldn’t be any surprise that even fewer people found the MP3 correctly — only about 32%! Somehow only the orchestral sample proved a little easier to guess correctly, with 48% of respondents nailing the MP3.
Conclusions
On average, only 44% of A/Bs were correctly guessed, crazy right? Well, not really crazy at all, actually! Can we really draw any kind of conclusion from this? Well, I think we can at least make a few observations. Firstly, for all the flak the MP3 format gets in the audiophile community, it’s actually not that bad of a format at all. Especially considering you can save 81-94% in file size compared to WAV (or 50-85% compared to FLAC). Secondly, however interesting these results are, these A/B comparison tests may not be the be-all-and-end-all of tests. Your mileage may vary, of course.
Interestingly, we did notice a few things parsing through the results of the test. For example: all audio engineers working with acoustic music, and most audio engineers in general, were able to discern between MP3 and WAV correctly and consistently. Audiophiles with a “high-end” set up had wildly varying results, however, ranging from 19% to 82% correct.
So should you just convert your entire album library to MP3 and save the space on your hard disk? Well, I personally think not. I’ve done the test myself — in a double-blind study with the help of my colleague — and the main difference between MP3 and WAV I found was that I just tend to enjoy the MP3s less, on a subconscious level. Something in me quite quickly said “nope, this is boring” when I was listening to the MP3 samples. That being said, as I have told everyone over the years: I’d rather listen to amazing music in a terrible MP3 quality, than boring music at hi-res PCM or DSD.
WAV or MP3: Most of you can’t hear the difference
It’s got to be one of the most talked about topics in the world of audiophilia: can you hear the difference between MP3 and WAV? We wanted to definitively know as well, so we conconcted a little test to find out, and many of you have participated, listening to the differences — if there are any. And well, turns out not many of you could hear any difference at all. And that’s no surprise.
How does MP3 compression work?
MP3 compression is based on a perceptual model of human hearing acknowledging that the human ear doesn’t equally perceive all frequencies. This leads to the possibility of removing some sounds that are inaudible, thereby saving space without noticeably degrading the perceived quality. The compression process can be broken down into a few steps:
The degree of compression can be adjusted. Higher compression results in smaller files but can affect the audio quality, introducing artifacts like pre-echoes or a loss of clarity in high frequencies. Lower compression ensures better sound quality but results in larger files.
The test
As briefly mentioned above, we concocted a little test to see if our listeners could perceive a difference between the original 44.1kHz 16bit lossless WAV and its MP3 counterpart. The test was divided into five levels corresponding to five compression settings in the MP3 decoder: Lowest, Low, Medium, High, and Highest Quality. At each level, we provided five samples from five different recordings we’ve made, in different genres and different instrumentations. Each track again has two samples, A and B, and it was your guess which one was the MP3-encoded version. So how did you score?
The results
Let’s start at Level 1 (lowest quality compression settings), and work our way up. About 55% of respondents couldn’t hear a difference between the MP3 file and the WAV file. Within this test, the groovy jazz sample with piano, drums and string quintet, apparently was the easiest one to compare, with 62% of you guessing correctly. The classical nonet recording turned out to be the most difficult, with only 52% of you picking out the MP3-encoded version.
At Level 2 (low quality compression), about 57% of you answered correctly, with again the jazz sample being the easiest (71%) and classical nonet the most difficult (43%).
At Level 3 (medium quality compression), the percentage of correct guesses which one was the MP3 drops significantly — only 36% of you guessed correctly. Here, the orchestral and the piano solo samples were the most difficult ones to hear the difference in.
At Level 4 (high quality compression), again only about 37% of you found the MP3 version within the A and B samples. The same samples again proved the most difficult ones.
At the final level, Level 5 (highest quality compression), it shouldn’t be any surprise that even fewer people found the MP3 correctly — only about 32%! Somehow only the orchestral sample proved a little easier to guess correctly, with 48% of respondents nailing the MP3.
Conclusions
On average, only 44% of A/Bs were correctly guessed, crazy right? Well, not really crazy at all, actually! Can we really draw any kind of conclusion from this? Well, I think we can at least make a few observations. Firstly, for all the flak the MP3 format gets in the audiophile community, it’s actually not that bad of a format at all. Especially considering you can save 81-94% in file size compared to WAV (or 50-85% compared to FLAC). Secondly, however interesting these results are, these A/B comparison tests may not be the be-all-and-end-all of tests. Your mileage may vary, of course.
Interestingly, we did notice a few things parsing through the results of the test. For example: all audio engineers working with acoustic music, and most audio engineers in general, were able to discern between MP3 and WAV correctly and consistently. Audiophiles with a “high-end” set up had wildly varying results, however, ranging from 19% to 82% correct.
So should you just convert your entire album library to MP3 and save the space on your hard disk? Well, I personally think not. I’ve done the test myself — in a double-blind study with the help of my colleague — and the main difference between MP3 and WAV I found was that I just tend to enjoy the MP3s less, on a subconscious level. Something in me quite quickly said “nope, this is boring” when I was listening to the MP3 samples. That being said, as I have told everyone over the years: I’d rather listen to amazing music in a terrible MP3 quality, than boring music at hi-res PCM or DSD.