An article in the Wall Street Journal relays the claim that changes to recordings to make them sound better when played from MP3 format causes the pre-MP3 original releases to sound worse.
"Right now, when you are done recording a track, the first thing the band does is to load it onto an iPod and give it a listen," said Alan Douches, who has worked with Fleetwood Mac and others. "Years ago, we might have checked the sound of a track on a Walkman, but no one believed that was the best it could sound. Today, young artists think MP3s are a high-quality medium and the iPod is state-of-the-art sound."
It isn't. Producers and engineers say there are many ways they might change a track to accommodate an iPod MP3. Sometimes, the changes are for the worse.
Is this claim true?
Eventually storage will become so cheap that compression of recordings will become less desirable. Also, newer formats will support more bits of resolution than CD offers and so especially at the lower frequencies sound quality should improve.
By Randall Parker at 2007 September 13 10:19 PM Comm Tech Society | TrackBackre: Eventually storage will become so cheap that compression of recordings will become less desirable. Also, newer formats will support more bits of resolution than CD offers and so especially at the lower frequencies sound quality should improve.
Could you explain why it would be the lower frequencies that improve? That doesn't compute in my foggy brain (I just got up). And today I don't have time to work through it.
You'd think that for professional-quality studio work that the sound engineers would just use a 24-bit or so PCM encoding with a 100kHz or so bit rate for each track as the raw recording because in the studio, the storage costs are trivial. Mixing this up could form the "master" file, also stored at hyper-audio resolution. The CD's 16-bit PCB@44.1kHz is nearly perfect for human hearing but when you mix zillions of tracks of classical music stored at this rate there's definitely a problem - although simple noise gating can clean it up a lot. It's a heck of a lot easier to mix and do audio transforms on PCM that it is in some of the more exotic formats like sigma-delta. Then for targeting into a compressed format like MP3, you'd just inverse transform the codec's artifacts.
There will always be a need for compression, as the limit used to be small storage (but NAND flash/Moore's law) eliminated that but there is still a need for audio and video compression for things like wireless because there is a harder limit on available RF spectrum. I also think people will just opt to store more and more stuff in their I-pods rather than go beyond 192bps MP3 because MP3 is good enough for our primitive primate ears. MP3 codecs often have some pre-image problems and this is apparent with certain things like clapping sounds. At 192kbps, you can hardly tell it's an MP3 even with the corner cases.
What I think would be really interesting is a future where microphones and 20-bit ADC converters are trivially expensive. You could "mic" a whole orchestra or movie set with a thousand microphones in a 3-d grid and perform all sorts of post-production acoustic imaging techniques that would filter out bad room acoustical effects, coughing, mistakes, etc...while making a plenoptic (actually plenacoustic) synthetic microphone with superior performance to any one physical one. It'd kind of be like the movie "The Conversation" only it would be trivial to do.
In the visual domain, you could also do this with zillions of cheap CCD cameras and do post-filming plenoptic tranforms to move the camera around to create new views and even filter out unwanted images like tourists, gawkers, or power lines. It could dramatically cut the costs of movie making since so much time and money is spent with expensive camera crews, lighting, and blocking off landmarks during filming. There's a very interesting PhD thesis from someone a the University of Washington that shows how to do this from a collage of still photographs. What's really cool about this work is that it visually shows the tricks and extends the concept of "filter" to something beyond math. Other tricks involve 3d model extraction from an array of camera images for automated video game making. There's still some work to be done in this area but one has to figure out some reason to fill up the $200 10TB hard drives that are only 5 years away.