[Music] So why did MP3 become the most popular format?

[Deleted]

didn't Apple introduced a DRM-less scheme for some iTunes songs?

iTunes music has not been DRM laden for years.

[Deviltry]

Lol, his cans are shit and his comparisons are even more shit.

His cans are shit, because - there's no such thing as good or decent PC speakers.

His comparisons are shit, because - no spectrum analyze, just "what I've heard".

Well I heard that you're shit, so what?..

And I bet his sound card is from Creative, that'd be MEGA fail.


Infracted. Express your opinion without the personal insults.

[Raphtheone]

So why did MP3 become the most favored format around the world?
My guess is probably the same reason QWERTY became the most popular keyboard layout.

What does mp3 have in common with the typewriter? Unless you're talking about a solution dating back to old technology which isn't being used anymore, I can sort of see your point. I consider myself among the average users when it comes to audio (with average equipment), so I don't listen to things above CD-quality (128kb/s?). This was the case when mp3 was released, so I don't expect mp3 to perform above it. As long as I can play my music in my music player, I don't care if it's mp3 or not, as long as the quality isn't too low (phone quality or similar). But I've got nothing against mp3. It performs its function well. Less can be said with some video codecs some people insist on using.

[Howard Moon]

Hi Xuvial, you might want to listen to what I have to say.

I work in the audio encoding industry, more specifically at the Fraunhofer IIS in Germany, co-inventors of the Mp3 and AAC formats, and developer of some of the world's best audio encoders. I want to answer to your question "why do people still use Mp3", and also to some shameless advertising of our product


About audio formats

If at all possible, please forget about Mp3. It is an obsolete format which is only still used because of brand-name recognition, and also because many mobile devices don't support AAC yet. AAC uses the MPEG-4 standard, which is much more advanced than it's predecessor the Mp3 (MPEG-2 layer 3 standard). If the devices you own support AAC (which often use the file extension .mp4), then there is no reason to keep using Mp3. I strongly believe that in the next 10 years, the average consumers will finally start switching over to AAC. And by then another more advanced format will have been developed, but hey that's how it works.

Keep in mind that AAC is a lossy encoder. Hardcore audiophiles claim that only lossless codecs such as FLAC are good enough for their trained ears. As an expert who does listening tests almost every day, I can tell you that even lossy formats have transparent quality at high bitrates. For the case of AAC, I consider 192 kBps completely transparent for most listening material. If you want to be really safe and have no storage constraints go with 256 kBps.

One thing to note about transparency is that at high bitrates, you will only hear small artifacts at critical sections in your music. For example, parts with really frequent transients (like applause or castanets) or high amplitude square waves (usually in electronic music but rarely by themselves). Nobody can tell between AAC@96kBps and AAC@320kBps unless it is a really critical item and they are using high-end headphones like Stax. Also, nobody can listen to a 5-second piece and say "oh this is AAC at 96kBps". They would need to do a comparative A-B test, preferably blind. In other words, at high bitrates quality differences only appear as isolated, sporadic artifacts, not long-term, constant disturbances.


What encoder to use?

This is the part where I advertise the Fraunhofer AAC encoder, since it's the one we develop Where can you get it? Easy: Winamp. Winamp used to contain the Dolby/CT encoder (our competition), but thanks to our quality improvements in the last few years AOL (owner of Winamp) decided to switch to the Fraunhofer encoder, and it is available with the free downloadable version of Winamp. Below are the settings I recommend:



Choose MPEG-4 AAC encoder. Then select constant bitrate (we could make a whole new thread about CBR vs. VBR, I personally prefer CBR) and AAC LC (low complexity). You don't want to use HE (high-efficiency), since that is optimized for low bitrates (32kbps and below) and uses SBR (spectral band replication) which has quite audible artifacts. Choose the bitrate you want but I suggest 192kBps.

As alternative I would recommend the Apple encoder (iTunes), but not the Nero encoder since it has scored lower than ours and Apple's in most listening tests that I've seen.


About resampling

I also saw that you mentioned having resampled to 48kHz. While higher sampling rates are better, resampling while encoding has no use. CD audio is recorded at 44.1kHz and as such, there is no energy in the signal above 22kHz (unless you are ripping Super Audio CDs, of course). Upsampling doesn't magically give you more highs than were present in the original recording. Your signal's spectrum gets extended to 24kHz, but there is no energy in the additional 2000Hz other than the noise floor.

Well I hope this helps, feel free to ask any questions regarding audio encoding. I earn my living with it

[Asera]

As alternative I would recommend the Apple encoder (iTunes)

I found when I tried converting my music library to AAC back when I had an ipod, it destroyed about 1/3 of the library on conversion - they weren't playable in anything.

[vesseblah]

Nobody can tell between AAC@96kBps and AAC@320kBps unless it is a really critical item and they are using high-end headphones like Stax. Also, nobody can listen to a 5-second piece and say "oh this is AAC at 96kBps". They would need to do a comparative A-B test, preferably blind. In other words, at high bitrates quality differences only appear as isolated, sporadic artifacts, not long-term, constant disturbances.

In pop and rock low bitrates are very easily noticeable, and not just with high end headphones and isolated events. Hi-hats and cymbals break with even modern encoders with below ~120kbps rates and on mp3 below 250kbps. Once you know the problem is there, you can't help to not hear it.

Keep in mind that AAC is a lossy encoder. Hardcore audiophiles claim that only lossless codecs such as FLAC are good enough for their trained ears. As an expert who does listening tests almost every day, I can tell you that even lossy formats have transparent quality at high bitrates. For the case of AAC, I consider 192 kBps completely transparent for most listening material. If you want to be really safe and have no storage constraints go with 256 kBps.

The reason why I use FLAC for all of my own use is archiving. I don't want to rip my CDs more than once, so I just rip them lossless. Discspace is really cheap (or was before the flooding in Thailand) and there's absolutely no need to use low bitrates for home use. Transcoding for portable devices is easy and fast from the original FLAC files, and when you're outside using some portable devices it doesn't really matter one bit since the traffic noise and poor quality amplifiers in the devices will hurt the sound quality more than low bitrates.

[squee666]

shall i be the 3rd poster in this thread to ask why you are not considering ogg vorbis.

[Jonuts]

If the testing wasn't double blind, then it is invalid. I've never been able to have anyone in a double blind tell the difference between a well encoded, high bitrate mp3, and an original file.

Except, you're leaving out half the test procedure. You do it with an ipod with stock earphones, I'd be *EXTREMELY* surprised if you could. Did you perform such a test using high quality audio equipment *AND* test using audiophiles, you know, the people that would actually pay attention during the test and note the differences? Just because a test is double blind doesn't mean the test was worth anything.

Anyways, onto the main subject, it's been long known that MP3 is an inferior format. You can get the same quality with more compression (And theoretically *BETTER* quality, but I can't tell ) using other formats. Of course, being a smaller file won't allow a codec to overthrow MP3. Data storage is cheap after all, and very very few people don't have enough room for MP3's. MP3 being basically a free format helps quite a bit too. You'd have to really put on your try hard pants to find a playback program that WON'T play MP3's. Not sure if such a beast exists...

[Howard Moon]

I found when I tried converting my music library to AAC back when I had an ipod, it destroyed about 1/3 of the library on conversion - they weren't playable in anything.

If you transcoded your files (converted your existing Mp3s to AAC) then you completely destroyed the quality of your audio. Encoding should be done once, and only once.

As for the compatibility of your files, not really sure why you weren't able to play them on anything. Maybe you had older/cheaper mobile devices which didn't support AAC (though an iPod should), or iTunes DRM prevented you from playing them (again, should still play on your iPod). Not sure, though, as I don't use iTunes, iPod or any of that iStuff.

---------- Post added 2011-11-14 at 02:18 AM ----------

In pop and rock low bitrates are very easily noticeable, and not just with high end headphones and isolated events. Hi-hats and cymbals break with even modern encoders with below ~120kbps rates and on mp3 below 250kbps. Once you know the problem is there, you can't help to not hear it.

Assuming the reference signal is sampled at 44.1kHz, AAC at 96kBps does no downsampling. This means that all 22.1kHz contained in the signal, including the highest pitched cymbals, are being encoded. The only scenario where you would hear coding artifacts, is when too many transients (i.e. cymbal hits) appear in one single frame (such as really quick sixteenths on a hi-hat). This causes the encoder to run out of bits for that frame, which leads to "holes" in the spectrum. Still, this would happen only very rarely and in very critical items, and would already be transparent at 192kBps. Unless of course the whole song is a continuous train of ultra-rapid impulses (some Aphex Twin stuff out there really kills encoders) you should hear no difference. As soon as the encoder needs to downsample, though, you will hear the loss of hi-hats that you described.

The reason why I use FLAC for all of my own use is archiving. I don't want to rip my CDs more than once, so I just rip them lossless. Discspace is really cheap (or was before the flooding in Thailand) and there's absolutely no need to use low bitrates for home use. Transcoding for portable devices is easy and fast from the original FLAC files, and when you're outside using some portable devices it doesn't really matter one bit since the traffic noise and poor quality amplifiers in the devices will hurt the sound quality more than low bitrates.

Archiving is a great reason for encoding lossless, and I agree that bitrate is the least of your concerns once you are listening to your music in the bus or train. But you bring a good point: before deciding how to to rip your music, you should know why you want to do it. What is the ultimate purpose? For storage/backup, for streaming, for the iPod?


One comment about OGG Vorbis. It is a pretty decent encoder, and it's patent-free. It scored just below the Nero encoder in the last big test that I saw. The problem in my experience has been lack of support from some devices, but that was a while ago.

[Howard Moon]

Jeeze, why do I always kill threads.. lol

[vesseblah]

One comment about OGG Vorbis. It is a pretty decent encoder, and it's patent-free. It scored just below the Nero encoder in the last big test that I saw. The problem in my experience has been lack of support from some devices, but that was a while ago.

MPEG-LA's patents are so damn vague and there are so damn many that I seriously doubt OGG Vorbis would live through dedicated attack from patent lawyers, just like Google's WebM (VP-8), or OGG Theora.

OGG Vorbis I think is too little and too late. It had it's moment somewhere between mp3 and aac, but failed to seize it due to lack of hardware support. There's no real niche for it anymore since AAC is "the devil you know" when it comes to legalities and beats Vorbis in the one area where it excelled over mp3... at very low bitrates which would have been ideal for portable devices.

[Rampant Rabbit]

I'm fine with listening to MP3's at 256 - 320 kb/s. Lossless formats only take up a lot of space for the small increase of detail you get.

Maybe if you have a professional monitoring system (which I have yet to buy) and trained ears you could hear the difference but otherwise it wouldn't really matter.


The reason MP3 is so famous is because of the small file size and the way it got accepted into the world. WMA came too late to take the leading role.

[Ravenblade]

When MP3 became standard (which was middle of the 90s) it was revolutionary. Compression rate, quality and being a streaming-capable format without carrying an expensive license tag made it quite popular. Of course it wasn't supposed to stay there forever but given enough lifetime and technological retention any format can survive decades before going finally obsolete forever. I remember quite well the beginning times of MP3, how everyone was praising it how *AWESOME* it is.

[vesseblah]

I remember quite well the beginning times of MP3, how everyone was praising it how *AWESOME* it is.

Slightly worse quality but half the size of mp2 files people used to pirate before mp3. On the downside mp3 files also required almost twice the CPU power to play back realtime which sucked pretty badly for people with 386's.

[Ravenblade]

Slightly worse quality but half the size of mp2 files people used to pirate before mp3. On the downside mp3 files also required almost twice the CPU power to play back realtime which sucked pretty badly for people with 386's.

Yeah, I remember setting up a SGI box in the data centre for decoding and audio-streaming to the aging SUN Sparcstation I was working at as it was completely overwhelmed by that. Also remembering how I set up my Amiga to decode and play back - even though it was powered by a MC68040 and already had a soundcard for 16 bit sound part, it was still quite occupied.

[Skippy88]

Since nobody touched on the "why" part it's because back in the infancy of the internet they were the first format that came out with low file sizes where you can easily download them from the internet. Looking at the guys test above MP3 was the in the middle of the 5 formats and the two that beat it were developed several years after MP3.

I still remember downloading my first songs from a guy I was talking to on IRC. Good times.