Apple Snow Leopard.. A cheap hi-fi upgrade?

[lovejoy]

...At least for Intel Mac owners anyway.

Having completely sold my soul to Steve Jobs in the last couple of years, I decided to shell out the £25 for an upgrade from Mac OSX Leopard to the new Snow Leopard today - for all the right reasons of course - improvements to the UI, increase in speed etc. and fair play, for £25 it was worth it just for that. It all runs a LOT quicker and just feels a lot more solid.

Reading into the changes however, I discovered something I hadn't realised.. Core Audio has been completely re-written. It seems that although pretty much everything looks the same, all of the changes have been made under the bonnet. So I thought.. As you do... I wonder if that's going to affect sound quality.

My conclusion after 4 hours of enjoying every note played out of my Macbook's optical port is "Yes it has done". With exactly the same setup as on my previous Leopard install - iTunes, Audio MIDI setup on 44.1/16-bit, optical into my Caiman these are my impressions of the changes:

Bass has become more taught, vocals have even more projection into the room. Instrument decay has improved: I loved the way that Joanna Newsom's used harp strings just seemed to hang in the air in front of you, quite breathtaking. When the music gets complicated and there's layers and layers of instruments, it's just more composed and never becomes mushy. It's more solid all round.

Something else I picked up on which was a surprise was the improvement in pace, phrasing and pitch. I became more aware of even the slightest deviations in pitch on a sustained note. There's also more absolute minute detail. You could probably hear a pin drop in an auditorium now.

I'd be really interested to know if anyone else has tried Snow Leopard and experienced improvements, or is it all just down to the fact that it's a clean install? I don't THINK it is as the old Leopard install wasn't that old and wasn't exactly clogged up with apps, but it would be good to get a second opinion.

Also, if you're pondering over Snow Leopard - just get it. For £25 it's a steal, and probably one of the cheapest hi-fi upgrades you'll get.

[Beechwoods]

Damn Apple for leaving the PowerPC in the weeds. My 64bit Dual 2Gb G5 is officially history. I have an Intel chipped titanium MacBook but I don't use it for music... I'd definitely be interested in whether your findings are universally held, Rich.

[Spur07]

Yeah, I'm also peeved with Apple for casting aside us G5 PowerPC users. I'll be extremely p*ssed if Core Audio has really improved in Snow Leopard. And all for £25. I experienced a massive upgrade when I went from Tiger to Leopard and USB II on my G5.

[Chippy_boy]

@lovejoy

When you think about this, if your OS is messing with the signal, then everything you say could easily be true. Maybe that is the case in your setup.

But equally - when you think about it - if the OS is merely streaming out bit-perfect audio, it is impossible for any of your perceived improvements to actually exist, isn't it. The bits are the same, and there is nothing in the new software that can improve the inherent jitter in the crystal in your Mac, nor the layout of the internal wiring - which also is the source of jitter.

There is no other way an sp-dif output can convey "information". It's either in the bits, or the timing of the bits. There isn't anything else!

I really do try to have an open mind about these things. But really, some claims are beyond plausible. Please understand - I don't doubt for one moment what you *heard*, I merely doubt whether such difference *exists*. i.e. I am not questioning your integrity at all, I want to be clear about that. And, as I said at the top, it may be that your software is/was not optimally configured.

EDIT: You might fing the following interesting reading. It desribes how to make sure you have your mac configured correctly and what can happen if it is not.

http://www.benchmarkmedia.com/wiki/i...enhancer.27.29

[lovejoy]

Hi Chippy_boy
Yes, I am well aware of the benchmark guides. I've pored through them many many times in the hope to get my Leopard based Mac sounding as good as my old XP/Foobar/ASIO setup, but never to any avail, so I'm well versed in how to set up a Mac properly for optimum sound.

You've proven my point quite nicely here as it happens, as Snow Leopard, with exactly the same setup DOES sound as good as the PC ever did.

The thing is, 'bit-perfect' is such a flawed phrase. There are so many things that claim 'bit-perfection' - for example, try comparing ASIO, WASAPI and Kernel Streaming on a Windows Vista system - all billed as bit-perfect, yet they all sound quite different. Bit perfect means absolutely nothing other than your original value between 0 and 65535 (on a 16-bit system) is still that value. There's no specification for how that value is dealt with by the OS - 16-bit values don't merely get passed, in-tact from your music player to the device driver for the sound card - they could be padded, reversed, encrypted, decimated, recalculated - any number of things, and that's before they've even got to the audio layer, which may perform another level of conversion so that it understands what's going on, because computers these days don't work in 16-bits so that data is not 'bit-perfect' between the output of iTunes and the audio IC - We now know that Core Audio has had a re-write, but even if the same code has been ported to Cocoa then timing is going to be affected, but it is not merely a port - I'm sure the details will emerge in time, but for now we're left guessing as to what they have done to it.

So the bit-perfect argument is flawed. It's exacly the same as saying all CD transports sound the same, and we know they don't.

But if that's not enough to convince you, seeing as you have an open mind about these things, have a listen for yourself. I could demonstrate to you an identical Mac audio setup, one using Leopard and one using Snow Leopard. I can guarantee you would not take long to notice the difference, as it is not small. I have not made this claim lightly without testing because I'll be the first to admit to being a victim of the placebo effect on more than one occasion, but this is marked.

Then, when you're convinced, maybe we can find out what is actually going on, rather than using the old 'bit-perfect' chestnut as a bludgeon, as the perpetrators of this argument are starting to look like zealots.

[twelvebears]

Hi Rich.

Very interested in anything which could make my Mac-based setup sound better and had been considering going for Snow Leopard anyway. For £25, why not!

Just not sure it would make any difference to me as I actually stream my audio via iTunes over a CAT6 network cable to an Apple TV, which is connected to my Beresford via optical cable.

In this situation, am I right in thinking that in my set up, the OS on the ATV is the 'weakest link' and therefore the potential improvements wouldn't apply?

[lovejoy]

Hi Steve,
Just to clarify, when you say you're streaming via iTunes over CAT6 - you're sharing the iTunes library on one machine with your ATV and then the ATV is doing the actual playback, is this correct?

That being the case, then yes, I would imagine it would make no difference as the underlying OS doing the actual music handling is still your ATV.

Even so, Snow Leopard is still worth the asking price as the Mac just feels so much nicer to use now. I'm beginning to sound like a fanboi... Oh dear.

[Chippy_boy]

Sorry Lovejoy - I don't mean to have touched a nerve, although it would seem I did.

Perhaps I am a zealot. On the other hand, perhaps I stand for reason and common sense.

All of your points about bits being padded and recalculated etc all sounds very plausible, but it's not actually correct. I am not saying that it never happens, merely that it doesn't always happen. Your description seems to imply that there are varying degrees of "bit-perfectness". Well this is not so. Perfect is a superlative. Something is either perfect or it is not. Whether or not a digital data stream is bit perfect, or not, is a boolean.

You can test it, very simply. Consider a DTS encoded track. You can imagine it as a sort of zip-file. The audio is mathematically compressed (lossily) and then decompressed at playback time. Like a zip-file, if the file is corrupted in some way, it can't be decompressed, in the same way you can't open a corrupted zip file.

You can try ripping dts-encoded cd tracks (they are hard to find but they do exist) into a lossless format such as itunes lossless. And then play them back through itunes into your HT decoder. If you get music, you know for sure every single bit is coming out as intended - they decoder could not decode it if the singnal was messed with in any destructive way.

I don't know if you can try this, but I have done so. In windows, kmixer.sys messes up the data. The only soluton in XP, as you may know is to use kernel streaming or ASIO, which is tricky (although not impossible) if you want to use Media Centre for example.

However, using a Mac - or in my case an Apple TV - you can and do get bit perfect output and the DTS encoded tracks come through untouched. As soon as you touch the Apple TV's volume control all is lost and you just get white noise.

So when I play tracks in bit-perfect mode, I do *know* they are indeed bit perfect. The data is absolutely identical to that which came off the CD, down to the very last bit.

Regards "timing", you mean jitter. There isn't any other sort of timing other than timing errors so gross that they would completely prevent playback altogether. As I said in my original post there is *nothing* software can do to change the nature of the hardware to reduce or to increase the jitter.

So, much though you might imagine how two pieces of software can sound different, so long as they are both streaming bit perfect, I still fail to see how it is technically - or indeed non-technically - possible for them to so do.

If you haven't got any lossless DTS tracks to test, I will happily upload one for you.

Cheers

[lovejoy]

Sorry if I came across as having a rant. You hadn't touched a nerve. I just know what I am hearing and unless you've heard the difference for yourself between Snow Leopard and Leopard then you're not really in a position to judge.

Yes, I know about the DTS test. Thing is, that does nothing more than prove the DTS signal is intact, so maybe that, to you and to all intents and purposes is bit perfect, but that doesn't have any bearing on the jitter/timing, however you want to call it.

All of your points about bits being padded and recalculated etc all sounds very plausible, but it's not actually correct

How do you know this? How long have you been working in software engineering for audio systems? Do you know how a signal is processed by your audio playback software, how the OS deals with the audio player, priorites, arbitration between tasks vying for processor time slots, data ordering and packeting, I could go on, and on and on, and then you begin to see just how many things can affect jitter on the way to a DAC. Of course software will make a difference.

So, much though you might imagine how two pieces of software can sound different, so long as they are both streaming bit perfect, I still fail to see how it is technically - or indeed non-technically - possible for them to so do.

..and this is the point that everyone who uses the phrase 'bit-perfect' seems to have such a hard time understanding. Now I am happy to accept that 'it just sounds better' because not being an Apple software engineer, the chances of me being able to analyse the code for myself to see what's going on are negligable. But the proof in the matter, as I have explained before is in two identical systems apart from OS, playing the same tracks, and sounding different. You can be as technical as you like about it, but your DTS test proves nothing more than the DTS flag is passing through the system unaltered.

Just because you don't understand how there can be a difference does not mean that there cannot be one. I would imagine that 99% of us don't understand how it works, but this is an audio forum, and I've started this thread because I'm finding Snow Leopard more musically enjoyable than Leopard or Tiger before it and I wanted to share my findings and see if anyone else shares those findings. So lets get back on thread and get a consensus.

[Chippy_boy]

I'm not off the thread. You are suggesting there is only one valid opinion, i.e. yours and I am suggesting an alternative possibility

i.e. that in fact the two OS's sound completely identical and that either something else is at play - different settings somewhere in the two OS's perhaps - or maybe your judgement may be flawed.

The "listen with your own ears" argument cuts no ice with me. Our ears - remarkable transducers though they are - are anything but reliable, given that are attached to, and interpreted by, the emotional lump that is our brain.

It's funny how that bottle of red wine tastes better on the balcony overlooking lake Como. But it's the same wine.

EDIT: In all seriousness I don't know the detail of what's going on inside the OS, but I did chuckle at your questions, given that I have worked in IT for 27 years and I used to write device drivers and have messed with OS kernels. But that was a long time ago, and I have forgotten more than I can remember!