Avrin

I've simply re-uploaded the decoder to MediaFire: http://www.mediafire.com/?jahzvgfgzow Just download and run the EXE (it's a self-extracting RAR archive with an installation script) - it unpacks two files, ATXDEC.AX and ATXPARSER.AX, to the C:\WINDOWS\SYSTEM32 folder, and then registers them in the system. Then you may use any standard Windows player to play non-DRM ATRAC3[plus] files, without installing SonicStage. The decoder doesn't support gapless playback though.

SimpleBurner is no longer updated or supported. The only way to use it with Vista/7 is from a virtual installation of Windows 2000 or XP.

Manually setting the recording level of the RH1 to 23 will maintain the original recording level. But the output of the RH1, even in the LINE OUT mode, is still not as loud as that of a Hi-Fi CD player. And remember that recording to PCM via optical in is totally useless, since the signal is changed during this. Transferring PCM from SonicStage, on the other hand, gives you a bit-perfect copy. If you are too careful, you may even use EAC to rip CDs, and then import the resulting WAV files to SonicStage and transfer - no $XX,XXX CD player will give you anything better.

1 Gb Hi-MD discs are less reliable than good old 80-minute ones. I've already had two FORMAT ERRORs on those.

Just discovered a funny thing about the Stereo USB. Its built-in application for playing music/viewing videos/artwork, plays music from the MP3 folder, and with gaps! And that despite the MP3s being encoded by LAME and containing all information required to decode them to a gapless stream (you may try that manually - the MP3s decompressed by LAME.EXE make a perfectly gapless set of WAVs).

The Beatles catalog had never been officially released on CD in decent quality (the only exception being the semi-official 1983 Toshiba Black Triangle issue of "Abbey Road" that was recalled soon after the release because of publishing rights not having been determined yet). The 1987 issues are not a state-of-the-art transfer. And there weren't any other serious releases before 09/09/09. Dr. Ebbetts's bootlegs made from vinyls are excellent, but still not sourced from the original tapes. While the stereo 09/09/09 remasters are a victim of the loudness wars (the mono remasters aren't, with the exception of the new "Mono Masters" discs), the end result is not as horrible as one may expect, especially in the 24-bit format. There isn't any clipping, and only some peaks are compressed by a few dB. Some quieter songs are not affected at all, and they actually sound better because of the higher resolution made possible by the level increase (the higher the level, the more bits are used to encode the signal).

The one I have is also 209 kb in size, and it also crashes WinAmp 5.57, even in its minimum configuration. So, we better stick with the Windows Media Player ATRAC3[plus] decoder I've posted several months ago.

Oh, man, the 24-bit version is better than the Stereo CD box. Just burned it to two DVD-Audio discs. Even compression doesn't kill the music. Low-level detail and shades of sound are excellent! Vinyl is finally dead. If it were not for compression, the release would be a sonic paradise.

As it turns out, the limited 30,000-copy release of the Beatles Stereo Remasters on an apple-shaped USB-drive contains the new stereo remasters in 44.1 kHz/24-bit quality. That is, supposedly better than CD. The problem with these is that they are also compressed. And why anyone may need compression on a 24-bit media is beyond me.

Rounding errors are introduced by any change of the digital signal (except doubling the amplitude in a decent program, if headroom allows). Avoiding clipping is obviously not enough. Level 23 in SONY portables keeps the original signal level without introducing rounding errors, but level 22 or 24 does introduce them.

The DAC in an external receiver may theoretically be better than that in a portable unit. But this in not always the case, since portable minidisc units are traditionally equipped with really good hardware. So you need an exceptionally good external DAC to make it worth the effort.

The RH910 is really great for classical music, and nice for other stuff.

Any manipulation with digital signal (including EQ, FFT, declicking and noise reduction) introduces rounding errors. If you still want to do this, first convert the signal to 32-bit in Audition, then process it the way you want, and convert the final version to 16-bit, using proper noise-shaping and dithering. This way rounding errors do not accumulate after each processing step, and the result will be much better. And never trust the internal dithering and noise-shaping algorithms used by Audition when processing 16-bit signals. Noone knows what these are, and it is always better to do everything in 32-bit, and use proper noise-shaping and dithering algorithms for the 32/16 conversion at the very last stage only, so possible errors do not accumulate. My approach to the 32/16 conversion: And don't forget that even PCM contains very little data for low-level signals in 16 bit format (lossy formats simply dump these), and these should be preserved to the fullest extent possible. It is the properly preserved low-level signals that make your recording sound really natural.

Yes. But don't forget about headroom. If the maximum peak level is below 50% - go ahead, but if it is seriously above 50%, leave the signal as is. Doubling the level does not actually increase quality, it only brings the signal on par with other louder recordings. Surely, if just a couple of peaks will get clipped, then this is also probably worth the effort (just don't forget to process the clipped peaks with hard limiting to make them look nicer - possible rounding errors for the maximum signal are negligible).

I'm not sure about hardware (although some should exist), but Adobe Audition perfectly doubles any signal, if it has enough headroom.

A lot of sound engineers fell prey to this delusion, and many good records are irreversibly damaged. +6 dB roughly equals 1.9952623149688796013524553967395 times, while 2 times roughly equals +6.0205999132796239042747778944899 dB. As you see, this delusion introduces a lot of rouding errors.

Just a couple of graphical illustrations. The first is a generated pure sinewave at the -68 dB level: It is already not very pleasant to look at, since only few bits are used to encode it, thus its shape is far from perfect. Now we add a 110% gain to it: As you see, the result actually looks (and sounds, believe me) much worse because of rounding errors.

The idea is simple. Only 16 bits are used to encode sound. And, out of these, even less are dedicated to low-level signals (say, 2 or 3 for really low-level stuff, and some 10-12 for louder). And if a signal is processed in any way, the results are rounded to the nearest whole bit, and this may seriously distort low-level components, which may actually be encoded by only 2 or 3 bits. Professional sound processing is done in 24 or 32 bits, and the final result is then converted to 16 bits using dithering and noise shaping algorithms, which, although they still add noise and distortion, make it sound "natural" and less irritating. But the 200% gain transformation, even when done in 16 bits, does not lead to rounding errors, since increasing the level 2 times is equal to shifting the signal by 1 whole bit - thus no rounding errors are introduced. You only need to check that the louder signals do not get clipped during the process (this never happens if the maximum peak of the initial signal is at or below the 50% level).

It is always a good idea to keep the original recording level. Changing the level of a digital signal (be it a CD or an MD) by an arbitrary value always leads to rounding errors and the accompanying distortion. Only if the original recording it really too quiet, you may double its level using some good program (e.g., Adobe Audition) without adding any distortion. But the change must be exactly 200% (not +6 dB), so as not to introduce any rounding errors.

Actually, SimpleBurner works with 4.3, but only under Windows 2000/XP.

Further studies revealed that the fake SP download mode is actually hard-coded in the main SonicStage executable, Omgjbox.exe. Address 0x00178DB7 (in version 4.3.01) contains the value corresponding to ATRAC3 132 kbit/s. And it looks like only ATRAC3 can be put there, i.e., it is possible to hack the program and set 105 or 66 kbit/s for fake SP, but other options, like ATRAC3plus or PCM, do not work.

It surely does. An optical out sends 44,1 kHz/16 bit Linear PCM, which is obtained from the original ATRAC[3] signal by the DSP in the unit. Then the receiver converts the Linear PCM signal to analog, and here the result strongly depends on dithering and noise shaping algorithms used in the DAC of the receiver, and on analolg filters employed after the DAC.

The authentication server is up right now.

Folk Alley

It's even better to use 32-bit Windows 2000 Professional SP4 on a Virtual Machine. It requires less resources and gives you no activation trouble. The list of system pre-requisites from Microsoft, for those who want to install SonicStage 4.3 "Ultimate" on Windows 2000 SP4 (some pre-requisites may already be installed, or can't be installed at all, but are still included for the sake of completion): Windows Installer 2.0 DirectX 9.0c Windows Media Format 9 Windows Media Format 9.5 Data Access Components 2.5