This is a time-frequency analysis of the FLAC signal of Can't Say Goodbye To Yesterday, from the album Metal Gear 20th Anniversary ~ Metal Gear Music Collection, released on Konami under the catalog number GFCA-53:
The file covers all the frequencies from the Compact Disc release for that album and song (for those that doesn't know, the Compact Disc standard.. the Red Book, follows the 16 bit / 44.1 KHz specs. so that means the max. frequency is 22050 Hz). On the picture, you can easily appreciate that the analysis detects sound from 0 Hz until more than 21000 Hz. That's the power of a lossless codification as FLAC, Monkey's Audio and many others, when getting the best of a source like Compact Disc.
Now lets take the MiniDisc standard in mind (credit notes that were submitted for this release declared this release was a promotional MiniDisc). The MiniDisc uses ATRAC as format (or Linear PCM too on the Hi-MD models). If I'm not wrong, taking them at their lowest resolution, the ATRAC on its ATRAC1 mode for example, gets also frequencies until 22050 Hz (through two stacked quadrature mirror filters which split the signal into 3 parts), and the Linear PCM goes to 8 bits / 44.1 KHz when talking about the poorest mode, so it gets also frequencies above 22 KHz. That's totally normal since those codifications and formats are focused on what human's ear listens, and those are sounds until frequencies like 20 KHz. Now you can see a time-frequency analysis of the 1st track os this album, Legend of Fantasy, from the FLAC file:
Frequencies above 16000 Hz are clearly killed, as normal on MP3 files. Now lets remember what our friend Myrkul shown us time ago.. if you all remember correctly, the 1st track was one of those at 128 kbps but the last one was for example the one at the higher bitrate (256 kbps). Lets see now a time-frequency analysis of the last song, Origin of Darkness, from the FLAC file:
URL of the IMG:
http://img145.imageshack.us/my.php?image=track8.jpg
(I don't know why this 3rd picture can't be shown on the post)
On it you can see richer parts. Don't be confused with the new amount of frequencies since that depends of the instruments from the song, the sounds that you can listen and so on.. but taking that apart, you can see more vivid colors on some parts too and that's simply because you've a little bit more of bitrate. Anyway, it's killed above 16000 Hz too in this case like any MP3 would be.
What means this?. This means that the FLAC codification posted on this thread is totally useless. It may be simply because they were done from the MP3 files posted above, or just because the original MiniDisck didn't include better audio than that (but for that, the only way is that the MiniDisc was including the audio only on MP3 through SonicStage, as a Hi-MD). I can't think of a promotional Nobuo Uematsu MiniDisc on that quality. Anyway...: credit notes said it was a Linear PCM Mini Disc recorded at 1411.2kbps, so no MP3 files on that MiniDisc. But now lets do a bit of maths:
If we've a Linear PCM recording of 1411.2 kbps, and we see it only gets until 16000 Hz (if I'm not wrong this is not usual for Linear PCM, but lets follow...) it means it has to be done at least at the double of Hz, so that's 36 KHz. If you have a recording of 36 KHz that gives 1411.2 kbps (equal to 1411200 bps), on Stereo (2 channels), you have the next: 1411.2 kbps = 36000 Hz * 2 channels * "?" bits. So, if you do that equation, you get that the "?" value is 19.6 bits. Linear PCM allows samples of 8 bit, 16 bit, 20 bit and 24 bit, and since it was -+ 16000 Hz in audio at the analysis, lets take this may be a recording of 20 bits / 36 KHz... but as I said.. this is not usual, and I can't think of Nobuo Uematsu using synths or instruments that produced sounds only until 16000 Hz. Just take a look at the time-frequency analysis and those cuts at 16000 Hz are produced like by the MPEG equation.
So, if we start from what's submitted to VGMdb on the Notes (MiniDisc at Linear PCM 1411.2 kbps), the only reason to see specs. like the ones we see at those screens on our FLAC files, is simply because they were done from a previous MP3 codification. I don't know if the FLAC files were done from the MP3 files posted above (a simple poor useless conversion then), and those at their time could be done from a proper Linear PCM MiniDisc. That may be possible, so the MiniDisc may still be real. But why on the earth someone do a MP3 codification from a Linear PCM 1411.2 kbps MiniDisc, at different MP3 bitrates.. (in the poorest ones, the 128 kbps ones, loosing more than 1200 kbps of sound) and later when asked, do a FLAC codification not from the original source, just from the already compressed audio files (MP3 ones)?
Excuse my grammar, I'm not english native. Just trying to clear out these things a bit more since we still don't see more answer to this thread. Any correction will be welcome. Full gallery of the time-frequency analysis at high resolution can be seen here:
http://img6.imageshack.us/gal.php?g=metalgearcantsay.png
