[joelgee]

I always download at 320 insane. I generally burn CDs and I find that to be the best alternative for me.
J

[uaburn]

I always download at wma Q98 VBR which is typically around 300-400Kbps.

The high bitrate option was the primary reason I started using aom3 and why I am still there... I am not an audiophile, but I can tell the difference if played on resonably equipment.

[RedSmurf]

Originally Posted by Asathor

I'm assuming I've mis-read your meaning above -- are you saying that you can take a 128 kbps mp3 file - and somehow convert that all the way up to a 320 kbps file? If so - then I believe you're unclear about the basic principles of lossy audio compression. All you've done by transcoding from the lower bit-rate is increase file size - but you can't re-introduce frequencies that have already been removed by the initial encoding.

Yes, by all means you should avoid transcoding (= decoding an MP3 and re-encoding it into something else). Transcoding a low-bitrate file into a higher bitrate will in the best case preserve all the artefacts of the low bitrate, but in most cases it will only make it sound worse. You can't just recreate the information that was thrown away while encoding at low bitrate!!! Transcoding a high-bitrate file to a lower-bitrate file is also bad, because it will cause more distortion than if you had directly encoded the original material into the lower bitrate. You can do it, but you'd need to use a higher bitrate than your 'target rate' to compensate for the extra distortion.

Of course, transcoding is exactly what Allofmp3 does with the Online Encoding feature. But when starting from 384kbps material as they do, the artefacts are so small that transcoding is perfectly acceptable. Of course, if you have the choice between OE and OEX, you should go for OEX.

I personally download in Ogg format, at quality setting 5 for 'average' songs, and 6 for the songs I really like.

[AlderaaN]

Originally Posted by RedSmurf

..Of course, transcoding is exactly what Allofmp3 does with the Online Encoding feature. But when starting from 384kbps material as they do, the artefacts are so small that transcoding is perfectly acceptable. Of course, if you have the choice between OE and OEEX, you should go for OEEX.

RedSmurf, thank you very much for putting the explanation together like this.

Although I'm well aware of the issues you've describe above, I've found myself logging into this thread 3 times since bkram's post trying to bring the technical info together, not even coming close to how you've put it.

Thanks man!

Samuel

[nappymonster]

I personally download in mp3, constant bit rate of 320kbps, unless the original is of a poor quality because it is old, in which case i download at 192kbps. I have a couple of questions though:

1) Which is higher quality mp3 @ 320kbps constant or variable?

2) Why would you wan't lossless?

3) If you did want lossless, which my understanding is "CD-Quality - exact copy", why are there different lossless? All are the same filesize, except PCM wave, which is twice the price and alot bigger file size.


Here are some facts about the song "The importance of being idle - Oasis"

Lowest price:
Ogg Q3 - $0.10 - 3.0MB
Highest Price:
PCM Wave - $1.01 - 39.2MB! For one Song - in this quality i could fill up my 4GB ipod nano with just ---90---- Songs. In 128AAC i can fit 1000!


Who would spend $1!!! per song?

Nappymonster

EDIT: Everyone who uses itunes does, but at a MUCH lower bitrate.

[nappymonster]

Oh, and where can i get this bit of software from?

Nappymonster

[RedSmurf]

Originally Posted by nappymonster

I personally download in mp3, constant bit rate of 320kbps, unless the original is of a poor quality because it is old, in which case i download at 192kbps. I have a couple of questions though:

1) Which is higher quality mp3 @ 320kbps constant or variable?

Variable would in most cases be better because it can temporarily allocate more bits to demanding parts in the song. For instance, if you would have a long silent part with not much going on, and then a wild drum solo, a quality-based encoder would spend few bits on the silent part and many on the drum solo.

2) Why would you wan't lossless?

For most end-user applications like listening to music on your iPod while commuting, lossless is indeed overkill. But if you want to remix the song and/or add effects, it's better to start from the original material than an encoding with artefacts.

3) If you did want lossless, which my understanding is "CD-Quality - exact copy", why are there different lossless? All are the same filesize, except PCM wave, which is twice the price and alot bigger file size.

I also find the existance of so many lossless formats a bit silly because in the end they all do the same. There will be some tiny differences in efficiency between the state-of-the-art lossless encoders, but they will all hit a hard bottom, being the information content in the audio stream. You can't go below that without going lossy(*). On average this means about a 50% reduction in size.
I think people prefer certain kinds of lossless encoders because of compatibility or ideological reasons. As for PCM, I think AOM3 only included that option either for incredibly lazy or dumb people, because you can get identically the same data for about half the size with one of the lossless encoders.

Who would spend $1!!! per song?

Nappymonster

EDIT: Everyone who uses itunes does, but at a MUCH lower bitrate.

A little known fact about the 128kbps (DRM) encoded songs in the iTunes store: they are all lowpass filtered to 15.5 kHz. This is probably to keep distortion within bounds, in the limited 128kbps bitrate. I didn't test the DRM-free songs yet, but since they have twice the bitrate, they shouldn't have the lowpass filtering.


(*) Theoretically, there are ways to go below the limit of the absolute information content in the song. But these methods would require the encoder to store an unpractical amount of sound fragments. In the end, the amount of data stored on the device that needs to be able to play the music, will still be the same.

[Nightfly]

Red Smurf..Thank you for your excellent post..I am much in need of the information you had to offer here...I have limited knowledge in this area
though I downoad every day for years..go figure.. If there is any other
info on downloading formats you would like to share ..I would appreciate it.
What do you really hear in difference on a portable between 192 and 320.. is their more definition?
I don't hear it..yes my ears are not what they used to be after all these years of loud music..
But I can only hear real differences on my home system that has a open sound with some sought of soundsatge.


Thanks again,

Nightfly

[RedSmurf]

Originally Posted by Nightfly

Red Smurf..Thank you for your excellent post..I am much in need of the information you had to offer here...I have limited knowledge in this area
though I downoad every day for years..go figure.. If there is any other
info on downloading formats you would like to share ..I would appreciate it.
What do you really hear in difference on a portable between 192 and 320.. is their more definition?

I have been playing with digital sound since we got our first Mac at home in 1989, which had 8-bit 22kHz stereo built-in

At very high bitrates, the difference with the original material will only amount to subtle noise. I've done many experiments with lossy encoded music, like subtracting the encoded sound from the original. What you get in that case, is a kind of 'ghost' version of the original track. It sounds vaguely the same, only much more silent and a lot uglier. This is because the idea behind all modern lossy codecs is that one can 'hide' noise under the actual signal, and the louder the original signal is, the more noise you can introduce without making it noticeable. This is due to the 'masking' properties of the human ear, you can read this wikipedia article to learn more about this. Masking is the reason why people have never really complained about the lousy dynamic range of vinyl records and audio cassettes: the noise and crackles could only be heard in silent parts of the music, and is masked away when the music is loud enough.

In practice, a lossy codec will quantize the frequency content of the music. Quantization causes noise, but the quantization is done in such a way that the noise is masked. The lower the bitrate, the fewer quantization steps and the more noise. When steadily decreasing the bitrate, at some point the noise will exceed the masking abilities of your ears, and the distortion will become audible. Because the psychoacoustic models used in lossy codecs are for an 'average human', that "ideal bitrate" differs between different people. The artefacts that are caused by the lossy compression differ between codecs and even encoders. At low bitrates, MP3 tends to produce 'ringing' or 'glass-like' sounds, while Ogg Vorbis will typically make the music sound more dull without introducing weird sounds.

The types of sounds that will degrade first when decreasing the bitrate, are sharp, random and sudden sounds like drums, cymbals and hihats. Pianos are also tough to encode because of their rich and subtle frequency spectrum. The same goes for singing. The most demanding sounds are from certain synthesizers because they have waveforms that couldn't even be produced by any natural source. There's a certain sound file on the web that will still sound bad when encoding it at 192kbps MP3 with the best encoder available.

The best way to determine what kind of bitrate and format is good for you, is to encode some 'demanding' songs (like with drums, piano, synthesizers) at various rates, and do some comparison tests. You can do these tests on your portable device and home stereo, and see what bitrate suffices for each scenario. Ideally, you should do a blind test, meaning that you don't know in advance to what bitrate you're listening. Otherwise your judgment may be biased.

[ibjim]

How much difference is there between aac and mp3? I think I've hit the best price to quality ratio with 320kbps aac with constant bitrate, but I've always wondered about variable at the transcoding quality? Would there be that much difference. It really annoys me if all my music isn't in the exact same format, or that I don't have sufficient quality to cover me for any future upgrades (I'm using some unimpressive Logitech 5.1's at the moment with the most airy sounding sub ever!).

The first time I realised that my previous choice of 192kbps mp3 wasn't good enough was when I got my Koss earphones. Cymbals were unacceptably compressed and tinny for my liking (I’m a drummer) and it really started to jar me! I go lossless when I can, and eventually get round to buying albums when I have the money, but for the most part I use allofmp3 as it’s the only place that has a wide range of music at an acceptable bitrate for a very reasonable price.

Why would I want to pay almost the full price of an album for 128kbps mp3’s? It makes no sense, I may as well just get the CD!