In relation to the standard of MP3, the bit price is especially vital – however not solely that. The software program used additionally performs a job. And this acquired higher over time. I attempted encoders from the final millennium and in contrast them with in the present day’s high quality.
Many individuals type an opinion on a subject as soon as after which by no means change it. One such opinion is, for instance: If you happen to can’t distinguish MP3 from uncompressed music, you will have listening to harm.
To place it this manner, it’s definitely fallacious. MP3 will not be the identical as MP3. At the beginning, it will depend on the bit price used. On the highest high quality degree of 320 kilobits per second (kbit/s), MP3 is taken into account clear. Which means that there is no such thing as a audible distinction to the uncompressed unique. This has been identified via blind research for a few years, if not many years. Funnily sufficient, it is truly the opposite approach round: solely folks with listening to harm can hear a distinction within the highest MP3 high quality.
In a blind check carried out by heise c’t, just one particular person may distinguish MP3 at 256 kBit/s from uncompressed music – and this particular person had impaired listening to. How is that doable? The reason is named the masking impact. When the bass could be very loud, we do not hear quiet mid-tones – they’re coated by the bass. Subsequently, they’re omitted from MP3 compression to scale back the quantity of knowledge. If somebody does not hear these basses or hears them a lot quieter, there is no such thing as a masking. Accordingly, this particular person hears the center tones – and in addition hears them in the event that they had been omitted from MP3.
The place does the opinion that MP3 typically causes ear most cancers come from? I believe from the final millennium. I bear in mind my first expertise with MP3 was destructive. This was in all probability primarily as a result of low bit charges that had been widespread on the time. 128 kBit/s was the very best of emotions, and that’s already within the vital vary. All the things under sounds noticeably completely different than uncompressed music.
The attention-grabbing factor about it: There have been additionally MP3s with 128 kBit/s, which sounded actually dangerous. This might have been as a result of MP3 encoders was once worse. And thus sealed the dangerous repute of MP3 as soon as and for all.
In search of a foul encoder
To seek out out whether or not my guess is right, I wish to create the identical MP3 with an previous and a brand new encoder after which examine them. The brand new encoder I take advantage of is a present model of LAME – it has been the usual for a very long time and is taken into account prime quality.
With the previous encoders I’ve to see what I can truly get to work. I’ve a Mac, and Apple does not imagine in backwards compatibility. After a couple of failed software program installations, I get my vintage Powerbook from 2002 out of the basement and create MP3 information on it utilizing iTunes. Like every thing on this machine, this takes what appears like an eternity.
Within the information created by iTunes, a lady’s voice will be heard in the beginning of every piece saying: “Encoded by N2MP3.” That is apparently the encoder my vintage iTunes makes use of.
However 2002 is outwardly already too fashionable – I do not hear any clear distinction in high quality in comparison with the present encoder. This at 128 kBit/s. I can not select a worse high quality.
Now I am trying round at Home windows software program – in any case, Home windows additionally runs on my M1 Mac. I rapidly uncover an attention-grabbing candidate for my experiment: AudioCatalyst 2.1. This CD ripper dates again to 1999 however nonetheless works on Home windows 11. He makes use of the MP3 encoder Xing, and it’s excellent for my challenge. As a result of Xing is taken into account to be very quick, but additionally unsatisfactory when it comes to high quality.
What I hear – and what I do not
Whether or not and the way nicely variations will be heard additionally will depend on the piece of music. I hear lack of high quality greatest within the excessive frequencies of drums or percussion. Or if there are massive variations in quantity throughout the piece. For this check I take advantage of 5 items that cowl the kinds of pop, rock, funk, jazz and classical.
To my shock, Xing generates completely usable MP3 information at 128 kBit/s. I can’t reliably distinguish them from the information that LAME produces. Within the blind check I suppose appropriately for the rock and funk items, fallacious for the classical piece and I merely do not hear any distinction for jazz.
Subsequently I scale back the standard to 64 kBit/s. No person needs to take heed to MP3s of this high quality, whatever the software program they had been produced with. However not less than this manner I ought to have the ability to hear variations between the encoders. The truth is, I can clearly establish the variations: The Xing encoder information constantly sound muffled – at 64 kBit/s, Xing merely cuts out the excessive frequencies to avoid wasting knowledge. The present encoder additionally has excessive frequencies in it. However that does not all the time sound higher. As a result of he has to avoid wasting the information elsewhere with the intention to obtain the extraordinarily low bit price. In consequence, the everyday MP3 artifacts are simpler to listen to.
The variations are so clear that I may even show them to you in a YouTube video – though YouTube additionally compresses the sound once more. The artifacts are significantly noticeable in an MP3 that I created by exporting from the GarageBand software program.
So I can say: Even with the identical bit price and the identical unique file, not all MP3s sound the identical. It will depend on the encoder. Nevertheless, this solely turns into actually audible at very low bit charges.
However that does not reply the unique query: Why did previous MP3s sound so dangerous? I’ve a couple of previous items by the band U2 from “unofficial distribution channels” which are encoded at 128 kBit/s. For copyright causes I can not present them to you, however I swear on my ear: they sound like shit. Which encoder had been they created with?
The command line software mp3guessenc analyzes MP3 information to seek out out. How dependable that’s – I do not know. However I will strive it.
For the U2 items in query, mp3guessenc specifies “Xing (previous)” because the encoder. Nevertheless, for the information I created with AudioCatlyst 2.1, mp3guessenc spits out “Xing (new)”.
1999 is definitely new for Xing. The software program was developed beginning in 1995 and offered to RealNetworks in 1999. There’s just one factor: I would like an excellent older encoder for my analysis.
The Xing encoder from 1997
So I obtain the Xing Encoder 3.0 from 1997. That is an 83 KB small command line software for Home windows. And there at the moment are truly clear variations audible. Most clearly within the pop piece “Take a look at This”. The drums sound duller all through your entire piece than in MP3 with a contemporary encoder. As well as, there’s a lack of dynamism and – fairly badly – there are sometimes even small dropouts.
Conclusion: Solely the very previous encoders are dangerous
An MP3 with the identical bitrate sounds completely different relying on which encoder it was created with. Nevertheless, on the bit price of 128 kBit/s I’ve to return to 1997 to seek out an encoder whose information sound clearly worse. Within the encoders examined from 1999 and 2002, the variations are so small that I can’t clearly assign them in a blind check. The variations are typically straightforward to listen to at 64 kBit/s – however that does not matter in apply as a result of no person hears in such poor high quality.
The selection of encoder might have performed a job on the very starting of the MP3 period. Nevertheless, there are a variety of different doable explanations for the typically substandard high quality of MP3s from file sharing platforms. The bit charges had been typically decrease than in the present day. MP3s weren’t all the time created from unique CDs both. I’ve burned CDs from MP3 information myself. When MP3s are ripped from such a CD, an already compressed file is compressed once more. The identical applies to music recorded from Web radio stations.
After this check, I’m additionally satisfied that MP3 has its dangerous repute from its early days. However the encoders had been in all probability a minor drawback.