Another thread on digital compression

There is a fascinating discussion here on the Harbeth site on psychoacoustics.

http://www.harbeth.co.uk/usergroup/...-and-audio-data-compression&p=10950#post10950

Alan Shaw is convinced that removing unnecessary data from the signal to the loudspeaker would reduce the strain on the transducer, and allow it to perform better where it does matter.

Others are disagreeing.

It would be ironic if he were right and tests could show a file from which data had been selectively removed could be consistently preferred in controlled subjective tests.

 

Brian, that is astonishing - i was just reading that thread over on HUG.

I suppose that logically, if you simplify the signal passing through the amplifier and speaker you make it easier for them to fulfil their function with less distortion.
However I'd be against the idea. I think the user should decide on the compression issue and not the speaker designer. Some users will demand full fat and other will be happy with semi-skimmed.

 

What is astonishing, Rob?

That I should reference the thread here while you were reading it there, or that Shaw should say what he said?

In the thread on a badged Harbeth Amp, he makes enigmatic reference to a "magic black box" - a plug-in. I imagined that would be DSP for room correction, but it could be on the fly data compression (stripping).

Some accuse him of living the the past - clearly he is mis-judged.

 

I'm not sure what's happening over at the Harbeth site recently, an over-zealous response to some questions has ruined the place and turned it into just another stilted manufacturer site.

On topic, I've not noticed my SHL5s struggling with full-fat 16/44 recordings.

 

The former, and that I was sitting here thinking that perhaps we could discuss the topic on ZG as you posted the thread

I like Alan and have much sympathy with his opinions. Also greatly enjoy the HUG though I agree with Basil that those running it could usefully lighten-up a little.

 

I'm listening to 24/44 on the old Rogers predecessor of the SHL5 at the moment and they also seem to cope extremely well.

 

This information appeared on another forum, but may be of interest to members here - JC.

 

Brian,

Interesting. I wonder if data compression a bit like engaging a rumble filter if you have a table and little speakers. Probably best that the speakers don't try to reproduce warps, lest the bass/mid driver go pfffft.

Joe

 

Of that thread, I find the BBC R&D department (RIP) document the most interesting. It's obvious when you think it through, but data compression + data compression = random nasties.

 

Excellent - thanks for that.

Last night broadcast this coming Saturday so I'll record that.
... at 16/44 wav in light of FN's comment above.

 

I agree entirely about not compressing lossy compression successively, but ALAC or AIFF are preferable to wav imo because of tagging. On the other hand if tags are not important, then wav is very universally employable.

JC.

 

320k mp3 should help your Tannoys out quite a bit, in warping to the current millennium, without breaking up, Joe

JC

 

JC,

What's the easiest way to get music stored on the iMac, which is in the office, to the Tannoys, which are in the man room?



Might be worth trying out this new technology, even though I have a decent CD player.

Joe

 

I'd go with a Squeezebox, assuming you have a decent wireless router.

Quite astounding sound for the price.


Yaaaaaa for me, 100 posts!

:MILD:

 

It makes some sort of sense.

In a non-linear device, as are all transducers (*), everything intermodulates with everything, leading to a wide-band layer of dirt added to the overall sound. Some of that dirt is bound to remain unmasked by the actual music.

So if the inaudible parts of the source signal are removed prior to hitting the non-linearity, there is simply less to intermodulate, and the net result might be overall superior.

This, of course, assumes then that intermodulation is one of the dominant distortion sources in the chain, and that the act of perceptually reducing the source signal, on its own, is not detrimental to the remaining signal. Perhaps a real challenge is posed here by the ever-varying nature of the compression. A signal part that is deemed inaudible in this temporal frame might become unmasked, and not fit for deletion, in the next frame. This goes unnoticed (by the very nature of perceptual coding), but what with any distortion products caused by this signal? They would switch off/on in time, and this might be problematic ...







(*) Maybe we should start pressing LPs from MP3 masters

 

Why don't you suggest that on another forum? It would raise a lively discussion.

 

I think the easiest way is to use an Airport Express, or a New AppleTv, either of which cost $99US+taxes.

This ease of use, is based on the fact that they are wireless and therefore don't require any cable connection.

The Aex has an analogue audio output on 3.5mm jack, and optical s/pdif from the same socket. So, to get going in a simple and low cost way you just connect it to the line-in on your analogue amps. Some people think the dac in the Aex is not as good sounding as an external dac (I'm open on this), so they use the s/pdif output with an external dac such as a Beresford, or a Benchmark, which then provides the analogue signal to the hifi amps line-in.

The New Atv has no analogue output so s/pdif into a separate dac is the way to go. I've not tried this product personally yet, but I shall buy one as soon as the local Apple store has supplies. They are available mail order now, I believe.

Both of these products assume you are going to use OS X and iTunes, and don't necessarily work with anything else. They use Apple Airplay to carry the signal wirelessly using standard 802.11 wireless ethernet protocols.

A software specialist supplier called Rogue Amoeba, makes an application called Airfoil, which can be used on Macs and PC's to transmit any audio to an Aex, from any application, which removes the OS X / iTunes requirement, and is very widely used by many people with great success.

Because an Aex is portable, you can move it around to any room, plug it in to the mains and get audio wherever you wish. An Aex will drive headphones if you wish, for casual or temporary use.

There remains the problem of how to select the required playlists if you cannot see the computer screen because you are listening in another room. This is usually solved by running a remote control app on an iPod touch, or an iPhone, which is functionally excellent, and gives you unobtrusive control of your iTunes library from your remote listening room.

Wireless alternatives are Sonos, or Squeezebox, but they are higher cost and not part of the supported Apple systems. For what it costs, I think Aex, or ATv, is a no-brainer.

If your CDP has an optical s/pdif input socket then you can use that as a DAC to provide the analogue signal to your hifi.

Alternatively to wireless you can use a cable. My choice would be standard ethernet cable which has no practical length limitation, but requires the use of a freestanding dac with an ethernet input. USB is limited to about 15 feet, and optical s/pdif about 30 feet, but co-ax s/pdif will go longer to about 60 feet but can be prone to induced electrical interference in long runs.

Finally Joe, I think you already have an iPod touch, so, a no-cost solution is to simply use the analogue out from the headphone socket to feed the line-in on your hifi. Most people think the sound quality from an iPod running ALAC is excellent for ordinary use. I suggest you try this first, if you haven't already done so.

Hope this helps a bit Regards, - JC

 

I don't have a deep understanding of how lossy compression works, but I am not convinced it would make the speakers job easier in any worthwhile way, for a good hi-fi.

If the primary method by which lossy compression works is to remove low level signal which would be masked by the main sounds, then they are naturally low level. If they are low enough in level to not be heard then they should be low enough to not really cause much audible distortion.

Think about it - if the signal that's removed is so small it can't be heard, then surely the distortion product caused by that signal must be even smaller, although perhaps not in the masking range I admit.

For non-hi-fi applications it might work, where the speaker is poor quality and stronger lossy compression can be used without it becoming negatively audible. Also I can see it might help in PA applications where very high SPL is required and the speakers are always pushed to the limit.

Also, I wonder if encoding in a lossy format always produce a less tasking waveform from a speakers point of view? It might be more suitable for file compression, but not easier for the speaker.

For example, a step function could be very easy to compress because it would go from '00000' to a sudden '11111111', very repetitive and easy to compress, however it isn't at all easy for a speaker to produce. A more complex wave with a gradual rise and a few wiggles on the way would be easier.

 

I think that whatever is outside the ability of human ears to detect might as well be removed from the signal, - after all if you can't hear it you don't need it, and you probably don't need the effects of it either.

Loudspeakers aren't 'ears' and will try to respond to all sorts of things that ears would reject. So will lots of other hifi components. (cartridges ??)

It's possible that signal processing is best applied directly after the microphone so that nothing that you can't hear ever gets into the signal ??

Werners suggestion to make LP's from mp3 masters, does seem to have merit.

JC.

 

JC,

I don't have an external DAC, so I think the easiest would be to try an Airport Express and use its line level out into the Stingray. The Apple TV thingy looks to be a neat product for 129 Canuck clams, but my TV is so old that it doesn't have any connectors on the back except for cable coax, and video and audio via RCAs, so it would be a wasted purchase. It's also ruled out for me because of the lack of an analogue output.

I've ripped a few CDs to the new iMac. It's running OS X 10.6.4 (or whatever the current version is) and iTunes 10, so I should be OK.

Is the volume controlled through iTunes?

I got a free 8GB iPod Touch with the computer, so I'm set. Is the remote control a built-in application or would I have to download it?

Nope. It's a Naim CD2, so no inputs of any kind and the only output is an analogue DIN connector.

Not ideal in my case. It would require running a cable down the hallway that connects the office to the man room and I've already used up my design concessions by having ginormous speakers.

I tried my Senns with the iPod and the sound was very good and certainly loud enough, something I didn't expect as the cans are 300 Ω and I think the iPod is expecting a much lower impedance. (I think 32 Ω or thereabouts is ideal.) I've also tried the Senns plugged directly into the iMac. Again, the sound was very good, but not at the level I get from the same CD being played on my CD player and amplified by a dedicated Creek headphone amp.

I don't know if the iPod and iMac are not quite up to running revealing 300 Ω cans, if the DAC in my CD player is better than in the ones in the iPod and iMac, or if the Naim/Creek combo is simply better but I prefer the Naim/Creek/Senn combo for critical listening.

However, I really can't fault the iPod with the Senns as a portable device. It's a very satisfying combo, but when I'm oot and aboot I'd rather risk losing the $15 earbuds Apple supplies than the Senns. (I got a great deal on them and doubt that'll happen again.)

Yes, very much so, and thanks.

Joe