AACplus/HE-AAC playback on Ipod 5G Video

[Chronon]

D'oh!  I feel silly now.    

Some of the files were off screen and didn't notice.  Thanks again, Llorean.

[mamboman]

Makes sense. I actually bought a pair of Sennheiser earbud headphones that output 112 dB (the highest I could find for earbud headphones), just because all the headphones I tried elsewhere couldn't handle the high volume (and the few that could still crackled and such). The pair I got are really nice, because they go into the ear, which helps cut out most outside noise and sends the music directly into the ear (quite enjoyable quality). I know it's crazy and such, but I got used to it years ago when my father would blast his older metal/hard rock music (this being back when I was like 5 years old).

that's crazy to listen at such high volumes, I am using a pair of Supferfi 5 Pros and I only listen with my iPod video at around the 20% volume settings. It's 119db sensitivity btw for the superfis i have.
I already have a little hearing damage on one side from listening to music too loudly since years ago, and i treasure whatever hearing i have now, i suggest you do likewise bro

[Davide-NYC]

Put "test_codec" in the "#if CONFIG_CODEC == SWCODEC" section of /apps/plugins/SOURCES instead of the "/* plugins common to all models */" section.

That way if you want to compile for a HWCODEC target (like compiling a sim of the archos) you won't get errors.  

[Chronon]

Thanks for the tip, Davide-NYC.  I'll keep that in mind for future use.  However, my current targets are all SWCODEC units and it compiled fine.  I haven't had time to run tests yet, though.

[TexasRockbox]

About 17 years ago I went to a concert by Eric Johnson at Austin's old City Colleseum.  Awful acoustics, no ear plugs.  One of the most excrutiating experiences of my listening life.  My ears rang for hours afterwards.  I  never went to another concert without using them.  The earplugs muffle the sound but it's worth not losing my hearing because I enjoy music too much.

For most serious listening I use some Stax electrostatics.  I've never actually tried to listen to any compressed music (MP3, OGG, ATRAC) using them.  I should give it a  try.  I'm quite pleased with the quality I'm getting with Q7 Ogg Apple 5G Video Rockbox.

The only AAC I'm familiar with is the ACC (Plus?) used by XM.    At their bitrate (96kbps?) if doesn't hold up for headphone listening to me.

What AAC setting is the general consensus for transparancy with Rockbox, if attempted on a Apple 5G?  Or is AAC just not recommeded at all?

[saratoga]

I would use MP3 or Ogg instead.  They're better supported, and functionally equivalent.  

[Gamesoul Master]

I like AAC because, at least in comparison to MP3, it provides better quality at the same bitrates (namely in the 100-150 kbps range that I use). I know little about OGG because of how little support it has with media players and media devices, so I've never bothered to even try it.

@TexasRockbox: AAC is MPEG-2/4 audio, the unofficial successor to MP3 audio (which is MPEG-1). It provides a quality superior to MP3 below 192 kbps (above that, it gets closer to equal). HE-AAC is High Efficiency AAC, which uses spectral band replication (and parametric stereo in v2, the newer but much less common version). XM uses the original HE-AAC v1, so you were indeed hearing HE-AAC music. If you didn't like that quality, you have very good ears (or they do some really horrible encoding). In that case... you'd probably not wanna touch anything encoded under a HE-AAC VBR below ~100 kbps (encoded as well as possible, which I'm sure XM doesn't do). If you tried LC-AAC (for listening on Rockbox or other devices), a VBR of around ~180 kbps with optimal encoding is good for transparency at decent volumes. These are both said under the assumption, of course, that you're transcoding/encoding from a loseless source.

[saratoga]

I like AAC because, at least in comparison to MP3, it provides better quality at the same bitrates (namely in the 100-150 kbps range that I use). I know little about OGG because of how little support it has with media players and media devices, so I've never bothered to even try it.

Well, thats what the marketing literature says anyway.  In practice theres very little difference, the most important change being an improvement to the licensing that allows for stores selling DRMed music.

[Febs]

In practice theres very little difference, the most important change being an improvement to the licensing that allows for stores selling DRMed music.

This test is now slightly dated, but it demonstrates Saratoga's point:



MP3 and AAC showed no statistical difference in this 128kbps test.

[Gamesoul Master]

Yeah, but that's just one test, and AAC has progressed since that time anyway. Look at this test or this one (that second one being by far the most recent), and it's clear that AAC beats out MP3, especially using good settings for the encoder. Creating the best MP3 and AAC files I could from a few songs, the difference was clear every time (which I did as a 96 kbps CBR blind test, since I wanted the best quality format without any bias).

But as anybody and everybody will say... it depends a lot on listening preference too. And so, test results will tend to vary from tester to tester.

For the record, I use Lame MP3 and Nero AAC, as those are the two best encoders I've found for those formats.

[Llorean]

Why wasn't LAME used for the 128 one?  It even clearly says that LAME is comparable at 128, and that the performance gap is solely because they ended up encoding at 112, basically meaning that your point of them being significantly different is said to be spurious even by the tests you personally cited. Considering they're comparing AAC+SBR to MP3 without SBR (there is a version of MP3 with it, though it's patented even more than standard MP3) it's rather a silly comparison anyway.

As well the first test you cited CLEARLY states "there was no winner in this listening test. Musepack didn't win. All modern codecs are tied at first place, simple as that." I don't know how you can misinterpret that. This puts Ogg/Vorbis sitting on level ground.

I'm not sure how the second test can state unequivocally that there's no margin of error in their results, and that their results are even VALID considering they send out test files to people who then listen to them on varying hardware.

If I send out 4000 test files from different formats, there's a possibility that everyone who gets the one from an Ogg/Vorbis will have worse audio hardware than those who get the MP3s, or some other silly combination. A proper test needs to be done on identical hardware by the same person, or the same set of persons each of them listening to every format. As far as I'm concerned that second test doesn't provide any useful results. Read their testing procedure and see for yourself, it's not even an ABX test.

[Gamesoul Master]

::Shrugs:: This is what they had to say about it:

Taking into account that Lame used less bits (112 kbit/s) for coding SoundExpert test samples, its performance gap doesn’t look so dramatic. Thanks to Lame developers MP3 quality is still comparable at the most popular bitrate – 128 kbit/s.

They had it set on quality 5... I guess the encoder itself decided to go that low for their samples. Still fairly valid, unless you think it could jump up 2 full points in 16 kbps...

As for the first test... reading that statement is all well and fine, and I'm not sure why he said it... but scroll down and look at the charts. Read the sentence or two of results for each one.

[Llorean]

I think it could easily jump into the "statistically equivalent" range, especially considering how flawed their entire testing procedure is.

[Gamesoul Master]

Possible. Though I really don't care... the initial point I was trying to make was that AAC is technically superior to MP3. That is fact. It may seem small to some people at this point, but that is due to the encoders, not the format. MP3 encoders have very little room left for improvement, while AAC encoders still have plenty of room. And during tests and discussions, it's quite rare to see anybody try to say Lame MP3 turns out better than Nero AAC (within the last year or two), but it's pretty common to hear that they're either roughly equal or AAC is superior (by whatever margin, it matters not).

Test results are irrelevant because of how much variety of music is out there. For instance, AAC completely dominates MP3 with metal music, but they are nearly tied when it comes to electronic music. In the end... it all comes down to user preference. And from my own tests, AAC suits my purposes much better than MP3.

[Llorean]

Yes, and you're perfectly welcome to your preference. But have you *any* proof that MP3 cannot be optimized further than AAC? Just because it's older doesn't mean it can't. AAC could've been made initially with less room for improvement. Have you done a thorough investigation of encoding methods and understand both algorithms well enough from an engineering perspective to be able to say this?

All we ask here is that you don't state something as "fact" unless you can actually back it up rather solidly. AAC is technically superior in the "it was invented more recently, and contains more modern techniques" sense, yes, but this doesn't mean that it will be better on average, or that MP3 can't or won't surpass it in the end. It's also inferior in some ways (resource requirements) so it even strongly depends on what grounds you're basing "technically superior."

Saying "I prefer AAC over MP3" is fine. Anything you want to present as objective better be backed up with facts that are provable.

And if you say "test results are irrelevant" then please don't cite tests to try to prove your points. Especially greatly flawed ones like that second one. It's rather hypocritical to attempt to prove a point with tests, then say they don't matter when someone points out how flawed they are.