When I think about it, I have 2 question in my head.

1. With today cheaper prize & higher perfomance in just about everything about technology and I see it will continue to be just that. Audio codec, which main purpose is to reduce it data size (Which is not really that large to begin with compare to something else like video) Will it get more develop & optimize further in the future? or it will just be "good enough"? And newer hardware&technology will render them not important.

2. If the first question is yes, it will still important and will be get further develop. Then, How far it can go? Because right now, I saw that today's codec are already good & developer seem to struggle to make it any better.So it's hard for me to imagine something revolutionary will going to happen. Like completely transparent lossy codec at 64kbps or lossless codec with 20% or more compression rate than today most efficient lossless codec.

Since I'm not codec developer or programmer, I can't or even try to answer my second question at all. So I hope someone with knowledge in this field will share his/her opinion.

Cheers.

I think that encoders still need to be worked on to become more efficient. I am talking about lossless encoders though. Lossy encoders will probably still continue to be developed on but their efficiency is pretty good right now. That said, some audio formats (such as AAC) still have plenty of room to grow where other audio formats (such as mp3) are nearing their limits due to the file format, not the encoder.

I think the development is in lossless encoders to make them much more efficient. I would one day like to have a lossless encoder produce file sizes that are similar to 320kbps mp3/AAC. That would be the day when lossless audio is truly portable. However it looks like developers are getting many reasons not to make extremely efficient lossless encoders. Storage space is drastically increasing and companies such as Apple aren't helping when it comes to portable audio. Why would a developer put in so much energy into making their lossless encoder produce extremely small file sizes whenever Apple has a 160GB iPod? That is why my idea will probably never happen. I thought it might happen as the market is switching to flash (SSD, NAND, etc.) memory but Apple released as 32GB iPod touch today. This shows that flash memory capacities are increasing just as much as hard drive capacities and that a player such as the iPod touch could very well be in the 64GB capacity range by this fall with a 128GB model released 6-12 months after.

So that is what I think. Developers could potentially further the efficiency of their encoders (particularly the lossless ones) but they probably won't given that storage space for both home computers and portable devices is increasing at extremely fast rates.

Question from a non-mathematician: is there any kind of ultimate limit to the compression a lossless encoder can achieve, for a given signal?

I'm thinking of the way that general purpose compression programs (Zip et al.) don't seem to have squeezed more out of typical text files in the last 20 years, and the way that differences in compression between present lossless formats are really pretty small (10% is not going to do much in the overall scheme of things, and people seem to have to work really smart to get that much of a difference).

As far as I'm concerned, there are three separate goals for audio codec development.

1) Decreasing power consumption for 'transparent' quality. This is of benefit for PC users (since the easiest way to decrease power consumption it to reduce file size) and portables users (longer battery life).

2) Smaller bitrates for 'radio' quality, so that internet radio stations can have more listeners with the same amount of bandwidth.

3) improved coding efficiency of lossless codecs.

Lossy encoders are already good enough at 128kbps. The last multiformat 128kbps test by Sebastian Mares proves this as well as the success of the Itunes Store. With increasing storage space, lower bitrates aren't as important as they were 5 years ago.

I agree with MichaelW. There is a limit to lossless compression and I think we're pretty near the limit. (Lossless developers please correct me if I'm wrong.)

I think optimizing encoding speed is also important, especially for lossy.

I agree especially with LAME. Wish it was as fast as the Vorbis Lancer builds.

I was just thinking this exact same question.


For nothing more than to say, "I did it." Also, I imagine with large archives of media it wont watter what kinds of storage space exist, there will always need to be some kind of compression algorithm.

I'm doubt this is it. My thought about why they don't do it is, They simply couldn't. With current method.

But this is just my guess of course.

But that does seem to assume that either they can't do it -- "simply couldn't" -- or they simply do it. And here "simply doing it" would seem to mean there'd be no real investment in time and/or money needed. But korngold2000 did ask "Why would a developer put in so much energy ...?" Software development isn't necessarily easy or cheap. At the extreme you have Mitch Kapor sinking millions of dollars in a software project, hiring dozens of the best-known names in the business, and -- years afterwards and completely disconnected from projected schedules -- having little to show for it. Stuff can be hard.

So I don't know either, and matters might not be as korngold2000 suggests, but it doesn't seem unreasonable per se to suggest that it might be possible but not easy to do something in software and hence that it might not be worth people's time.

Solid state HDD's
Interesting topic, I think people will continue to work on it; from all angles.
The smaller a file can be, the faster it can be transferred and the more economically it can be accessed.
Although anyone can say "why bother when theres a 160gig HDD"

Why did anyone make any format at all after CD's? CD's are lovely. At the time, I'm sure there would have been people saying "this is it; the absolute pinnacle of audio in both sound and affordability and economy"

Likewise it's hard to estimate ANYTHING with anything to do with computers since there is always such rapid improvements. Even though that was taken into account the original idea relies on everything advancing.

But like I said above, people will work at it from all possible angles. By the time someone is able to download a full length lossless album in a millisecond mankind will not cease to improve on things. Computers etc only get more advanced and more powerful, but likewise corporations only create and employ more and more controlling limitations to the stuff at what seems to be an equal rate. But thats another subject

Some information:
MPEG is working on a new coding technology. It focused at low bitrates (24 kbit/s per channel and lower) for mixed content (music and/or speech). It will have better quality than HE-AAC v2 and AMR-WB+
http://www.chiariglione.org/mpeg/hot_news.htm
http://www.chiariglione.org/mpeg/working_d...AC/USAC-CfP.zip

Depends on the market you're referring to. In the consumer arena, MP3 has pretty much won the battle on the lossy side of things, and while further development is possible, anything that breaks compatibility with the spec probably won't happen. As the limit of development for the spec is reached, it's possible that "good enough" may just be fine. After all, with storage costs plummeting, the incentive for small file size is not nearly as great as it used to be several years ago.

Well, let's see.

In 1976, someone scoffed at the idea that 32kb/s speech coding would be necessary. In 1979, someone else scoffed at the idea that 16 kb/s sbc speech coding would ever be useful. In 1986, the idea that someone would want to do music at 4 bits/sample was regarded as a joke, and as completely unnecessary. In 1989 the MPEG committee wanted to deep-6 Layer 3 because nobody would use it. In 1994, they wanted to deep-six AAC because nobody would ever use it. Obvious, nobody uses any of MP3, AAC, or WMA-Pro nowdays, do they, and we don't have increasing network congestion at all. Do we?

One example of not successful new image compression technology:
jpeg2000
Results of testing jpeg vs jpeg2000:
http://compression.ru/video/codec_comparis...parison_en.html
http://compression.ru/video/codec_comparis...mparison_en.pdf

jpeg2000 is far better than jpeg but neither one browser (Windows OS hasn't it) has native support of it without additional setup of plugin. jpeg2000 was developed 8 years ago.


Another example of successful new video compression technology: MPEG4 H.264 AVC standard is getting popular very fast.
Hardware support : HD-DVD/Blu Ray, mobile, game devices Apple etc...
Software support : last Flash player, Apple Quicktime, etc...
There are already rumors about new wavelet video compression H.265 that will come approx in 2-3 years.
And this year company on2 is planning to release their new VP8 alternative codec to compete with h.264/h.265.
Video compression is on its hottest period. Audio?

I think there's a bit too much hype around wavelets. The I-Frame coding of H.264 is supposedly already better than what JPEG2000 does -- at least for higher rates. (Sorry, no reference. Woodinville?)


I think audio compression is already quite advanced today. Video bitrates of H264 streams are half of what you need for the same quality when MPEG2 is in use. In the 128-ish area MPEG4-AAC is not much better than good ol' MP3. On the video compression field there's possibly still more room for improvement simply because of the high sampling rates (pixels/sec) of uncompression video. Due to processing power constraints video codecs tended to be of low complexity in the past and I guess with more processing power that's available today and tomorrow you could throw some more advanced techniques at it like some sort of lapped transform (not necessarily wavelets) as replacement for the in-loop deblocking filter, more advanced ways of predicting motion or repeating textures etc etc ...

The latest improvements for audio coding focused on really low bitrates. Parametric coding tools have been added to MPEG4 audio. But these hardly improve the sound quality when used for moderate to high bitrates.

just my 2 cents,
SG

Yes. It isn't confirmed whether wavelet transform is more efficient than (M)DCT or any other effective lapped one. There are some claims about dirac wavelet based codec. I saw some video 3 years ago. Dirac vs x264. It seemed to me I frames of Dirac had better quality. Maybe it was effect of lower quantization of I frames compared to others types of frames. Yes, quality wasn't constant and it's hard to make any conclusion of that. Currently there is no stable wavelet codec to compare.

But there is some discussion about next generation H.265 and wavelets in area of video compression.

I'm probably quite disconnected from audio lately, as I was not even aware of this.
It seems to me that the timing for proposals is quite tight, so there will likely only be proposals by companies which already have something right now.

My guess on proposals:
*FhG
*Dolby
*FT/"Orange"

and perhaps Ms