Regarding the Audio VCD with 7 hours of "CD quality" music...

[redeemer-dk]

Let's say the CD is 650 MB.

The AudioVCD contains MPEG Audio, which is MPEG-1. I'm not sure whether Layer II or Layer III is used, which will also have an effect on the quality.

VCDEasy's site claims that it can hold 6 hours of "CD QUALITY MUSIC" (and doom9 claims it can hold 7)

That is: 6 * 3600 = 21600 seconds

650MB equals (650 * 1024 * 1024) / 1000 * 8 = 5452595 kbit.

5452595 kbit / 21600 seconds gives a bitrate of 224 kbit/s (the result is 252 kbit but that is not a valid bitrate), not to be picky, but 224 kbit mp3s are not CD quality, and if it's Layer II this makes it even worse. I could have accepted this on VCDEasy's site, but not Doom9's site. That's the same as saying DivX is DVD quality (i hate when people claim that)

Just little rant :)

-Redeemer

[adam]

First off, (S)VCDs are burnt in mode 2, which means that a 74 min cdr could hold 740MB's worth of data, rather than 650MB's. Also, without even reading any of the claims, I wouldn't be suprised if "cd" was referring to an 80 min cdr, which holds 800MB's in mode 2, rather than a 74min cdr. I don't know about you, but where I live I can't even purchase 74 min cdrs anymore.

Now (S)VCDs only support mpeg1 audio layer II audio, not layer III, so no that's not ideal. (Yes SVCDs do support mpeg2 audio, but that's another story.) Its true there is a lot of contention as to what bitrate mp2 and mp3 become "cd quality," (or if that is even possible)but most sources do list those points at around 224kbits and ~128-160kbits respectively. Many audiophiles would surely disagree with those numbers, but that does seem to be the general consensus.

Also, if you were to make a SVCD audio only disk instead of VCD, then you could use higher bitrates, ie: anything up to 384kbits.

[redeemer-dk]

But in order to burn in mode 2, isn't it a must to have some kind of error protection built into the format. MPEG Audio doesen't have that.

And the fact the it infact IS layer II makes the quality worse. I've read several places that CD quality (or close) is around 256/384kbps MP3, so i would claim that MP2 224kbit isn't.

And if you do burn at 384kbit which may be cd quality, then it can't hold 6 hours, which was the claim.

[KpeX]

1. Quality is subjective and not directly based on bitrate. for example, quality settings such as those used for vorbis, AAC in some cases or lame --alt-preset will generate many different bitrates based on the audio source/genre.

2. MP2 and MP3 are lossy codecs, tough to compare them with uncompressed cd audio. I'm not sure I see the point of this thread.

[int 21h]

Mode 2 is used where data integrity is not crucial (VCD, SVCD, etc.), it is not a requirement for the data itself to have any sort of error protection.

[RadicalEd]

Originally posted by redeemer-dk

And the fact the it infact IS layer II makes the quality worse. I've read several places that CD quality (or close) is around 256/384kbps MP3, so i would claim that MP2 224kbit isn't.

Actually, layer II does better at higher bitrates than mp3 does, so I wouldn't doubt 224 - 384 being decent enough to call CD quality. Then again, CD quality is a stupid term, sounds-like-a-cd is more like it.

[redeemer-dk]

RadicalEd: Actually, layer II does better at higher bitrates than mp3 does, so I wouldn't doubt 224 - 384 being decent enough to call CD quality. Then again, CD quality is a stupid term, sounds-like-a-cd is more like it.

Where have you heard that? I doubt that is true since Layer III is an expansion of Layer II which was made for programs where the complexity should't be too advanced. I agree that CD quality is a stupid term, what i mean is the quality that sounds like the original CD, not the same exactly.



KpeX: 1. Quality is subjective and not directly based on bitrate. for example, quality settings such as those used for vorbis, AAC in some cases or lame --alt-preset will generate many different bitrates based on the audio source/genre.

I know that the audible quality varies between the complexity of the audio track, and that Vorbis and AAC performs better at the same bitrate as MP2, but we're not talking about them we're talking about MP2 :)




int 21h: Mode 2 is used where data integrity is not crucial (VCD, SVCD, etc.), it is not a requirement for the data itself to have any sort of error protection.

Won't that create problems when reading the disc? I'm not sure actually how sensitive a CD is and how often the data recovery protection actually IS used, but if a certain byte can't be read on the disc, it could cause many strange effect, especially in crucial places of the bitstream.


-Redeemer

[int 21h]

Originally posted by redeemer-dk
...
Won't that create problems when reading the disc? I'm not sure actually how sensitive a CD is and how often the data recovery protection actually IS used, but if a certain byte can't be read on the disc, it could cause many strange effect, especially in crucial places of the bitstream.
...

The strange effect is that the audio would skip or stop playing for a moment. Read up on MPEG and you'll see that the decoder is made to be very resilient. Usually in MPEG the error is not corrected, but skipped altogether.

[redeemer-dk]

I haven't actually read the coding of the audio data, only frames etc. But still i would think that a skip/stop in the audio would be unacceptable.

[int 21h]

A skip or stop in the audio is not defined as a catastrophic error though. An error from which the decoder cannot recover (i.e. the DVD player locks up, the decoding program crashes, etc.) is catastrophic and the standard does its best to reduce/eliminate those.

[KpeX]

I know that the audible quality varies between the complexity of the audio track, and that Vorbis and AAC performs better at the same bitrate as MP2, but we're not talking about them we're talking about MP2

I'm aware of that. The point was that you can't say 'x bitrate = cd quality' or any other quality for all audio.

[fccHandler]

Originally posted by RadicalEd
Actually, layer II does better at higher bitrates than mp3 does, so I wouldn't doubt 224 - 384 being decent enough to call CD quality.
I have to agree with that, because it reminds me of a similar argument that MPEG-2 is better at high bitrates than MPEG-4 (which I also agree with). Here's my 2 cents: The more complex Layer III compression method eats away and destroys more of the original source data to achieve its greater compression ratio, which the much simpler Layer II doesn't (and can't) do. Therefore, at very high bitrates (say, 320+) I think that a Layer II compressed version will be a closer reproduction of the original sound than Layer III.

[int 21h]

You have to wonder though, with the same quant matrices, the same bitrate, resolution, etc., how different will Mpeg-4 look from Mpeg-2?

[fccHandler]

Good point. But even when those things are equal, I suspect at some deeper level MPEG-4 is throwing away more data to achieve more compression. Granted, I really don't know as much about MPEG-4 compression as I do about MPEG-2. All I can say is at very high bitrates (say above 5000 Kbits/sec) MPEG-2 starts to look better than MPEG-4, at least to me. ;)

[int 21h]

I think by default, Mpeg-4 applies too much quantization to certain luminance values. (i.e. black never looks like black, to me at least..) Perhaps if you used the same matrix though, maybe it would look better. There are probably XviD people who know.