The iTunes Mp3 codec is old very old has not been updated in ages and it's always failed to Lame
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple's Lossless Audio Codec (ALAC) Now Open Source
- Thread starter MacRumors
- Start date
- Sort by reaction score
The iTunes Mp3 codec is old very old has not been updated in ages and it's always failed to Lame
Have you even used the iTunes encoder since 2001?
Either that or LAME has degraded since its release.
It has NOT FAILED TO LAME SINCE 2008.
It has been UPDATED TO BE AS GOOD AS LAME.
It has BEEN ON PAR WITH LAME SINCE THE TIME OF THOSE TESTS.
Or you're just trolling and need to really get your head out of your ***
If you say so
This is the only Mp3 encoder you should be using http://wiki.hydrogenaudio.org/index.php?title=LAME
Did you even read the comments of that thread you posted. Sorry mate you're wrong.
In the mean time pm me i have a great price on some snake oil for you.
Last edited:
I'm just going to quote this post, because I'm grunching the last few pages of this thread, and I just want to focus on the issue of compressed audio somewhat, as there was some fairly heated discussion going on here for a while about how "bad" lossy compression really is.
Well, I was somewhat curious about exactly how good or bad different compression types and bit rates were, and also to a certain extent how much my hearing is capable of ascertaining any differences between them.
I'll provide more details briefly, but in a nutshell - I was absolutely gobsmacked by how GOOD certain lossy compression formats have become. I will say here and now for the record that I have an incredibly hard time distinguishing between the FLAC version of a number of high quality recordings and a, wait for it....... 80kbps HE-AAC of the same programme material.
These tests were not scientific in nature, but simply borne out of curiousity to see what I could find perfectly listenable with no noticeable loss in quality on my daily listening rig, which consists of an admittedly modest setup of Ultrasone Proline 750 and Sennheiser HD25-13 headphones, running from a Fiio E9/E7 headphone amp/DAC combo. I have yet to perform this test through my Metric Halo ULN2/Adam P11a combo, but I shall do that in due course.
I took a number of subject tracks - some that I listen to regularly and some known for their high recording values (Chesky and B&W for example), and spent a deal of time encoding each version into a variety of formats using XLD.
I encoded using Ogg with values ranging from 0-10 in steps of 2 and found that I stopped being able to detect discernible differences from the source at a value of somewhere close to 4, and by the time I hit a value of 6 the lossy version was effectively transparent to me.
I encoded using Lame MP3 VBR with values from v8 to v0, and while Lame often suffered more than other formats on most material, it was often perfectly acceptable for daily listening from v2 upward (upward meaning a lower number in this instance)
I also tried AAC, using both CBR and VBR encodings, and surprised myself by how often both 128kbps CBR and 96kbps VBR were so close to the source material as to prove indistinguishable for me, with my ability to differentiate or perceive any differences totally disappearing on this equipment at bitrates any higher - and the 256kbps AACs were certainly way past what I would have considered to be "transparent".
However, the biggest shock of all for me was just how low I could get the bitrate to go when using HE-AAC and still have a FANTASTIC sounding file. an 80kbps HE-AAC file generally reduces down to <10% of the size of the FLAC that I encoded it from, with an average file coming in at somewhere between 2-3mb, but with such a negligible loss in quality when compared to the FLAC that in regular daily listening, where I am listening for the enjoyment of music rather than critical analysis of encoding formats, I would be more than happy to have my library stored in this format.
I have now concluded that for my own purposes, having access to lossless source files is still important for various reasons, but for day-to-day listening in non-critical environments on anything but pure audiophile equipment, I would have absolutely no problem whatsoever with almost my entire music library being encoded with HE-AAC at a bitrate of 80kbps.
If it turns out that my hearing is just bad and I can't discern some obvious differences, then so be it, but just to try to put my experiences into some kind of context, I have spent numerous years of my adult life as a professional recording engineer, and I have also taught recording engineering at college level. Whilst at 37 my hearing may not be what it was in my 20s, for what it's worth I believe I still have reasonably good hearing. Although this isn't the most relevant metric, my pitch perception is still very sharp, as according to the test over at http://tonometric.com/adaptivepitch/ I am still reliably able to differentiate two tones 0.5625 hz apart at 500hz, which they classify as exceptional. So, I may not have "golden ears" but by the same token my hearing and audio awareness is still far from bad.
80kbps ...... I'll be damned.
Last edited:
Interesting. Can I ask what software you used to do the conversions during your testing?
What source music did you use?
I've selected LAME v0 as my compressed standard for iTunes (I rip as ALAC and keep those files too). I have a test file that I always try out, and I can clearly hear differences between LAME v0 and ALAC - but it's an 'edge case' with this one track, and I haven't found anything else that's quite as clear. I'd like to try this track with HE-AAC to see how good it really is.
FYI- The track is 'Overkill' off the (same named) 1979 Motorhead album. It may be counter-intuitive, but it's the very fact that it's not a 'clean' recording that makes compression so difficult. All the way through there's a lot of splashy hi-hats and natural reverb - and I can hear compression 'ducking' artefacts even with very high bitrate LAME encodings.
I certainly don't doubt your findings. I last did comparative testing several years back when I selected v0 - that was before HE-AAC became standardised, and I'd consider moving to a newer format if it was beneficial.
When you say clearly how was this tested under a double blind listening test? Like using Foobars ABX component. The vast majority of people fail and the few others have a difficult time to the point it's not worth worrying about because out of ABX testing it would be virtually impossible to hear the difference.
If you can clearly hear the difference between V0 and lossless then go sign up to HydrogenAudio forums and you could be a great help to the Lame project.
Why one must conduct a double blind test otherwise claims are meaning less http://www.hydrogenaudio.org/forums/index.php?showtopic=16295
If you can clearly hear the difference between V0 and lossless then go sign up to HydrogenAudio forums and you could be a great help to the Lame project.
Why one must conduct a double blind test otherwise claims are meaning less http://www.hydrogenaudio.org/forums/index.php?showtopic=16295
Last edited:
Conversion software was XLD. It's important to note with XLD that if you choose the standard "HE-AAC" option from the dropdown list of codecs it will max out at 64kbps, which at times can suffer from a couple more problems than 80kbps in my limited testing. To achieve 80kbps you need to select "mpeg-4 aac", and then in the options for that codec select the "use HE-AAC checkbox, which will automatically cap the usable bitrate at 80, or if it doesn't, just select 80 for the bitrate from the drop-down list. All these options are found under the "General" tab of the software's Preferences.
Some of the test tracks were:
"52nd street" from David Chesky's "The Body Acoustic"
Allan Taylor's "Colour to the Moon" from "B&W Audiophile Recordings vol. 1"
"Come On Up To The House" by Sarah Jarosz
"Violin Concerto #1 In A Minor, BWV 1041 - 1.Allegro Moderato" as performed by Anne-Sophie Mutter from her album of Bach's Violin Concertos (this has a harpsichord accompaniment which is useful for testing smearing etc - lots of low bitrate encodes tended to get really messed up by this and the harpsichord almost turns into a sitar, but Ogg did really well and so did HE-AAC)
"Across this Antheap" by XTC
Last edited:
Thanks, I've just been trying that - and yes, initial results seem really good even at 80. It's late here now, I'll spend more time trying tracks tomorrow (and when I can properly turn up the volume).
My test track selection is usually informed by how poorly a track compresses under ALAC (Overkill is 1077kbps). Interestingly, my worst-compressing albums under ALAC all seem to be by the Japanese Electronic artist Ami Suzuki - the recordings sound very bright and I think they've used a lot of Aphex Exciter to give them extra high harmonics and 'bite'.
I'm mostly listening on a MOTU D/A and Genelec 8020 near field monitors.
I initially did comparisons 3 or 4 years back using a Benchmark DAC1 which is the most brutally revealing piece of equipment I've ever owned. I got rid of it, since it was a lot of cash for something I didn't use a whole lot. I'll be interested to see if I can hear differences today, using my current setup.
Of course, I wasn't using double blind either - so there will be some other influences.
Last edited:
The iTunes MP3 encoder isn't just on par with Fraunhofer, it IS Fraunhofer.
Apple paid to license it and use it.
Thanks for the info, just checked the "About iTunes" menu today and found that out too
Point being, there is no difference between iTunes encoder and most popular encoders out there. The ABX test between lossy vs lossless is valid.
Thanks for the info, just checked the "About iTunes" menu today and found that out too
Point being, there is no difference between iTunes encoder and most popular encoders out there. The ABX test between lossy vs lossless is valid.
You are being the stubborn one.
...if you want a fair test, use the LAME codec. iTunes uses an inferior mp3 encoder. Don't believe me? Try it for yourself The results are startling, even at 320kbps. iTunes mp3 encoder is based off of the Fraunhofer mp3 encoder it was modified heavily for speed and has not been updated in a very long time.
iTunes as an Mp3 encoder is inferior.
Last edited:
Nuetral Source of info.
Is there an article or site that explains the various formats?
I would like to learn more before trying to find the version that fits my ears the best.
I do not need a video of someone playing Zeppelin backwards.
I get the idea that this is all done to make files smaller and the various ways to do this each take some sort of data out?
I can do without some data I'm sure, but how do I know what I am missing?
What is the lowest and the highest bit rate?
I still believe this is a good move by Apple.
Is there an article or site that explains the various formats?
I would like to learn more before trying to find the version that fits my ears the best.
I do not need a video of someone playing Zeppelin backwards.
I get the idea that this is all done to make files smaller and the various ways to do this each take some sort of data out?
I can do without some data I'm sure, but how do I know what I am missing?
What is the lowest and the highest bit rate?
I still believe this is a good move by Apple.
For lossy encoding, yes. Unless you're limited by storage space (e.g. on an iPod), why bother when you can use lossless and never worry about what you're missing.
Fun little experiment; encode a song in a lossless format (WAV, Apple Lossless etc) and encode the same song in a lossy format (AAC, MP3 etc). Then import both tracks in to an audio editor (e.g. Audacity) and invert one of the tracks. Then you'll be able to hear what you're missing when you encode with a lossy codec.
For mp3 and AAC, 320kbps. CDs are 1440kbps (uncompressed), but compressed they can be around 600 to 1100kbps.
Pro-Res is a perfect example of how letting others use your codec increases adoption...but it's a bit late in the day. What would be better is to allow formats on the ipod...but Apple isn't exactly about choice is it?
Who even cares about mp3 anymore. It's an old format and inferior to modern codecs such as mp4/AAC at any specific bitrate. You might as well be debating which coal is best for running a train, or which gas is best for lighting your house.
1. Go to iTunes
2. Open iTunes menu
3. Click "About iTunes"
4. Wait 4 seconds
5. Stare in amazement at the message "MPEG Layer-3 audio coding technology licensed from Fraunhofer IIS and THOMSON multimedia."
Which means iTunes DOES USE Fraunhofer as its codec.
Last edited:
I can confirm that on the source material I tested, 80kbps HE-AAC sounds noticeably better than both Lame v0 and Fraunhofer 320kbps VBR, both of which have big artefacting issues on low level noise details, which the much lower bitrate HE-AAC doesn't suffer from, and this is by far the most audiblly distracting of encoding artefacts to my ears at least. I consider it a huge bonus that the HE-AAC file comes in at about 25% of the size of either MP3.
Note these tests were simply for personal preference, not for any scientific comparison.
Just wanted to say thanks for the last few posts. Extremely informative, and really nicely written and detailed. Same to @firestarter for the questions and additional conversation.
I'm starting to convert our CD library over, and have been thinking about some optimal, but "casual" listening formats/CODECs (the "master" digital file will of course be lossless, though recently, I've been thinking ALAC vs. FLAC).
Sorry to totally derail the thread, but what's your take on Audio Technica AD700s for an extra set of phones (non-Amped)? If you'd prefer to answer via PM (if at all
), that would be outstanding too.The ones that want maximum compatibility care about mp3, its one of the only formats the fit across a variety of devices. In fact there are only a handful of players that will support AAC.
That being said, I prefer AAC over MP3..specifically at the lower end of the spectrum. 128kbps AAC is vastly superior to MP3 at the same bit rate. I think above 192kbps, however, the differences are very little.
Just wanted to say thanks for the last few posts. Extremely informative, and really nicely written and detailed. Same to @firestarter for the questions and additional conversation.
I'm starting to convert our CD library over, and have been thinking about some optimal, but "casual" listening formats/CODECs (the "master" digital file will of course be lossless, though recently, I've been thinking ALAC vs. FLAC).
Sorry to totally derail the thread, but what's your take on Audio Technica AD700s for an extra set of phones (non-Amped)? If you'd prefer to answer via PM (if at all
You're very welcome - I hope they are of some value to you and save you the time that I spent comparing the various formats
Unfortunately I have no experience with the AD700s, but if you haven't done so already, a search over on www.head-fi.org is likely to yield a number of reviews and opinions on them. Might be worth checking out if you have yet to do so.
----------
After the little exercise I conducted last night, I would now be inclined to disagree with you. The problem with *any* MP3 codec at *any* bitrate is that (to my ears, at least), it does a terrible job of handling low level noise detail, so if your music collection has any tracks that contain quiet passages with a perceptible noise floor (for example, tape hiss), then an MP3 conversion will render this noise into a swirling, garbled mush that sounds like what I can only describe as white noise being flushed down a storm drain
As far as I can tell, none of the AAC-based codecs at any bitrate suffer from this incredibly distracting artefacting issue. Your results may vary depending on the type of music you listen to, but for me this really is a deal breaker.
Last edited:
You're either ignorant on the subject or one of those vinyl worshipers that think it's a magic format because it contains even-order distortion (i.e. 'euphonic' distortion since the ear tends to like even-order distortion) and/or because early compact discs were poorly mastered.
The key test is if you record the output of a high-end vinyl rig to digital and play it back with a double-blind test whether one can tell the difference or not. The answer to that is an unequivocal and resounding No, you cannot tell the difference.
I'm replying because I subscribe to high-end crap 'theory' for some time and then I got a degree and a job that allowed me to buy high-end equipment and discover for myself the difference between FACT and WISHFUL THINKING. A given recording and mastering can go either way, but the CD format is VASTLY SUPERIOR IN EVERY WAY to vinyl and anyone who thinks otherwise is kidding themselves. One only need do double blind testing to prove it beyond argument.
I'd love to see one iota of evidence through double blind testing that 256kbps AAC is NOT audibly transparent. I've never seen any data that was better than statistical chance. I used to converse online with one of the engineers that was a part of the team that developed AAC and believe me, they went to great lengths to test that format for utter transparency at higher bit-rates and very good sound at lower ones that other formats would sound like total crap in.
But for any topic on audio quality, the analog, vinyl and 24-bit+ audio proponents tend to show up and spread their opinions like they're proven facts.
You can use any quality of equipment you like for a double-blind test. Stereophile was often challenged to tell the difference on various items (given the snake-oil their magazine pushed for the sake of advertising dollars) and they had/have access to some of the most expensive equipment around. They couldn't even tell a $30k tube amp from a TFM modified $500 amp Bob Carver made. Needless to say, they poo-pooed double blind testing as 'inadequate' the last time I looked (understandable given their advertisers would abandon them if they admitted they were the same as cheaper equipment).
I certainly use much better quality equipment than computer speakers. I have Carver AL-III ribbons bi-amped (with about 500 watts of power RMS) with a custom active crossover in my upstairs listening room. Those use the same ribbons that the Genesis II used (they licensed/bought them from Carver at the time) that sold for $50k a pair (Carver charged $2k a pair for them). Stereophile LOVED the Genesis speakers, but somehow didn't like Carver ones. They couldn't explain why. (The answer was in advertising dollars, I'm sure).
Downstairs, my home theater uses matched driver speakers from PSB in a 6.1 configuration (identical 3 speakers across front underneath a 93" projection screen). The PSB satellites in question are rated +/- 1dB for frequency response. The room is digitally corrected for frequency aberrations in the room response itself. You might have guessed by now that I don't like crappy sound quality. Even my computer speakers in front of me are Klipsch 2.1 sub/sat (best 'computer' speaker ever made, IMO). I also use a pair of them in my bedroom (whole house audio enabled).
I use JVC noise-cancelling headphones at work and on airplanes (pretty good sound quality for the most part, apart from some odd effects from the noise-canceling once in awhile). I have Koss Studio headphones at home for music production (I use Logic Pro on my MBP). I have a 6-speaker/sub setup in my car (with USB and iPhone enabled head-unit). I record in 24/48 for headroom, but there is no need for that in playback given the actual dynamic range and frequencies involved.
I don't listen to crap audio. There's a difference between quality high fidelity audio reproduction and elitist nonsense.