RSS feedToday Engadget posted a rumour (from AppleInsider) that the next iPod would have direct audio input, using a built in MPEG-4 or AAC encoder. If this was true, then I’d seriously consider this the MiniDisc killer, which would be amusing considering every one of the couple of hundred MP3 players released in the last 12 months were supposedly iPod killers.
A little history of MiniDisc is probably in order. Back when Sony and dutch company Philips invented CDs, we suddenly had digital audio in our lounge rooms, cars and even in our Walkmans. This was a great money spinner for Sony and Philips, not because they could sell CDs, as Sony wasn’t actually in the music business at that stage, but so they could sell their manufacturing plant technology and the compact disc certification mark to the consumer electronics companies. Only Sony and Philips had developed the CD manufacturing technology, electronics company were required to license the playback LASER technology from them, and the record companies were required to pay for the privilege of having that little compact disc logo on their product. This is why the current CD DRM technologies which prevent digital copying of CDs has Philips a little frustrated and Sony in a bit of a schizophrenic quandary, because the DRM doesn’t actual conform to the Sony and Philips standard, and therefore cannot use the compact disc logo, which ultimately means they don’t have to pay for it either. Sony of course is now in the record business, having bought Columbia Records back in 1989. Also, you can tell the difference between the Philips and Sony manufacturing, due to the see through plastic centre on a CD, which is clear for Sony and opaque for Philips. But I digress.
The problem with CDs, and why people were still buying cassettes, was because the CD was read only, and home equipment that could manufacture a CD seemed a long way away, until of course Pioneer invented the technology to do it. At least my memory says it was Pioneer, so I may be wrong. In fact I searched the CDR FAQ and I couldn’t find a reference to it, but I’m sure if you email the maintainer, Andy McFadden, who is also an old Apple IIer like me by the way, he’ll track down the answer for you.
So to plug the gap, Philips invented the Digital Compact Cassette (or DCC), a digital version of the old stereo cassettes we knew and loved, which made sense, considering they had also originally invented the cassette to begin with. Digital audio, in it’s raw form, is simply a series of values representing the position of a waveform over time, in the case of CD, 44100 samples per second at 16 bit resolution. 44100 or 44.1KHz was chosen because the maximum frequency that our ears can hear is around 22KHz or so, and 44.1KHz gives you at least two samples per wavelength, at that frequency, which should represent as close enough to the positive and negative amplitudes of the wave for playback. The original Fairlight music computer sampled at 50KHz by the way, and DAT tape, while variable, is able to sample at 48KHz, which is why DAT is still so popular. These samples are called PCM or Pulse Code Modulation, and are the basis of digital audio. Anyway, in order to store the huge amount of data required to store digital audio, Philips came up with a technology called PASC or Precision Adaptive Subband Coding. The basic idea was that you chop the incoming audio into a dozen or so frequency (or subband) bands, ranging from low bass sounds up to high 22KHz, removing sounds which probably can’t be heard from each band, and then joining them back together again. This effectively compresses the data, but it is of course lossy, so every time you record with it, you’ve lost data from the original waveform. However this was fine, because by designing it to effectively be good for only one generation of copying, you have a built in DRM. The problems with PASC, were that the bands were divided equidistantly across the spectrum, whereas sound is inherently logarithmic. This meant that the lower bands actually represented more perceptive range than the upper bands. Perhaps this was supposed to address the compression of harmonics problem, but I’ll come to that a little later. Anyway, DCC failed. It wasn’t random access, so you still had to fast forward and rewind, the PASC obviously wasn’t ideal for home taping, and the audio head was still using metallic particles oriented via magnetic forces (the same as in standard cassette tapes), to simulate a purely digital recording format. With these type of recording heads, like in standard cassette tapes, the tape itself rubs against the head, causing both the tape and the head to wear down. But the big reason it failed, was because of Sony.
Sony came up with MiniDisc, using a similar analog head rubbing against metallic particles technology, but developed it as a rotating disc, giving them random access capability like a CD. They also developed LASER based guidance for accuracy, so that the analog read/write head could use more closely spaced tracks and subsequently store more data. Sony also developed their own compression scheme called ATRAC or Adaptive TRansform Acoustic Coding, which works similar to PASC, but divides the signal into 52 logarithmically divided subbands instead, giving each band equal importantance in the spectrum of hearing. Having killed off DCC, Sony is still flogging this 1980s based technology as modern audio equipment.
The big flaw in PASC and ATRAC is the fact that sound, particularly in music, is based on harmonics. A simple note played on a guitar for example, such as A, which is 440Hz, isn’t just 440Hz, it also generates harmonics at doubling intervals, so 880Hz, 1.7KHz, 3.5KHz etc. The problem is that these harmonics fall into different subbands when compressed, and may or may not be removed if the compression feels like removing them. So, pull out a couple of harmonics, and you end up with a more echoey or thin presence of the sound. This is the basics of why MP3 and the rest is so crap at low sampling rates. The importance of harmonics tends to be lost on technologists, which is why audiophiles still love vinyl, and a lot of professional recording is still done in the analog domain.
MiniDisc and DCC use lossy compression, same as MP3, AAC and MPEG-4, they’ll all degrade through successive generations of copying. That’s why the record companies aren’t completely up in arms about this, because most audio luddites will rip music at some really low encoding speed and resolution, which makes it sound tinny and echoey, and won’t realise how bad it sounds. A recent article by Jupiter Research claimed that with personal devices, particularly MP3 players, increasing their storage, there was limit at which people would want probably no more 1000 songs, and therefore was just increasing memory size for the sake of publicity. What they fail to realise is that increased disk storage actually means the capability to finally return to raw non-lossy PCM encoding for much higher quality audio. I can finally toss that 1MB song away, and have a perfect digital copy at around 60MB instead. As bandwidth and storage increase, lossy compression such as MP3 will become a distant memory and a short 20 year period in history, which we’ll look back on with melancholy.
Now, where was I? I don’t believe I’ve remembered all this crap. Oh yes, the new iPod, the MiniDisc killer. This needs a new post.