A paper by Audirvana developer Damien Plisson
http://www.amr-audio.co.uk/large_image/MAC OSX audio players & Integer Mode.pdf
http://www.amr-audio.co.uk/large_image/MAC OSX audio players & Integer Mode.pdf
A look inside OSX audio
- Thread starter Vincent Kars
- Start date
Steve Williams
Site Founder, Site Co-Owner, Administrator
Thanks again Vincent, but this is pretty standard audiophile player fare, justifying the author's player. Nothing new here that I saw. "Optimization," isolation from noise, and timing not slaved to the computer. Theory, no data. Not even any jitter measurements comparing Audirvana to iTunes, much less anything establishing the audibility of the dragons being slain. I remain convinced that jitter is the harmonic distortion of the new age: A technical differentiator crawling around in the shadow cast by a lack of practical data, imagining benefits that remain unsubstantiated.
Still, there appears to be a free demo. I've given Amarra and Pure Music a shot. Maybe I'll try this one too.
I have usb connected to an outboard box that isolates and re-clocks, but it's there for the convenience of multiple outputs to headphone and speaker systems. I've unplugged it and sent optical directly from the MacBook to the DACs. That doesn't change the sound at all. What does is my mood. Wish someone would come up with a jitter reduction scheme for that.
Tim
Still, there appears to be a free demo. I've given Amarra and Pure Music a shot. Maybe I'll try this one too.
I have usb connected to an outboard box that isolates and re-clocks, but it's there for the convenience of multiple outputs to headphone and speaker systems. I've unplugged it and sent optical directly from the MacBook to the DACs. That doesn't change the sound at all. What does is my mood. Wish someone would come up with a jitter reduction scheme for that.
Tim
The Audirvana player software is free. There is no "demo", because you can keep using it forever. Damien accepts donations, but does not require them. I think it's a great way for folks to try out computer audio with a dedicated player on a Mac without shelling out any more $$$ to get started. He deserves kudos.
Lee
Lee
Hi Tim
Remember the early days at CA when all you had to do was
- Get a Mac
- Press play?
This oasis has gone.
Industry has discovered the Mac as an audio platform and as a consequence you get all kind of claims.
Our player uses HOG mode!
Our player uses INTEGER playback!
For only $900 this benefits will be bestowed upon you.
That’s what I like about this paper.
It simply describes how it works.
Hog mode= exclusive access to the sound card
Integer= bypassing the OSX mixer.
Pretty much like WASAPI in exclusive mode in Win.
Compared with the hype as generated by Amarra I appreciate his plain explanation of how OSX audio works and what possibly might improve the sound.
Whether it does, only your own critical ears will tell.
Remember the early days at CA when all you had to do was
- Get a Mac
- Press play?
This oasis has gone.
Industry has discovered the Mac as an audio platform and as a consequence you get all kind of claims.
Our player uses HOG mode!
Our player uses INTEGER playback!
For only $900 this benefits will be bestowed upon you.
That’s what I like about this paper.
It simply describes how it works.
Hog mode= exclusive access to the sound card
Integer= bypassing the OSX mixer.
Pretty much like WASAPI in exclusive mode in Win.
Compared with the hype as generated by Amarra I appreciate his plain explanation of how OSX audio works and what possibly might improve the sound.
Whether it does, only your own critical ears will tell.
It is a very well written paper. Just enough technical data but not too much.
There is one flaw in the paper and general thinking in this area:
"Minimizing synchronous CPU load taken for the audio data streaming operations. In addition to reduce jitter, this also helps to reduce audible RF interferences patterns, especially in low frequencies."
Jitter is *variations* in timing data. If the timing difference is constant, then it becomes delay which is harmless. So what we are after if we want to reduce jitter is to reduce variations, not absolute differences.
In that sense, minimizing CPU load does nothing and may actually make things worse. Let's look at an example.
Case 1: CPU load at 90% all the time. Since the load is constant, so is the variations in the system. Therefore timing differences will likely stay constant on digital connections.
Case 1: CPU load is at 5% on the average, but it consists of spikes 1000 times a second followed by no activity. Guess what? You will have now have a 1000 Hz jitter on your output! That's because it is the sudden change that causes the output waveform to change.
Of course, it is possible that even case 1 causes variations as there is no way to actually have a constant load. The system is running at billions of operations/second. How on earth do we characterize its load? Think: the timing counter on your player is causing CPU activity as often as it is updated.
So in that sense, PCs are disadvantaged here and one has a hard time putting much value on such assertions. It is best to side-step the problem altogether, not use the PC for timing at all using an synchronous, and isolated device which has a tiny microprocessor in it and not subject to such large variations and complexity as the PC.
There is one flaw in the paper and general thinking in this area:
"Minimizing synchronous CPU load taken for the audio data streaming operations. In addition to reduce jitter, this also helps to reduce audible RF interferences patterns, especially in low frequencies."
Jitter is *variations* in timing data. If the timing difference is constant, then it becomes delay which is harmless. So what we are after if we want to reduce jitter is to reduce variations, not absolute differences.
In that sense, minimizing CPU load does nothing and may actually make things worse. Let's look at an example.
Case 1: CPU load at 90% all the time. Since the load is constant, so is the variations in the system. Therefore timing differences will likely stay constant on digital connections.
Case 1: CPU load is at 5% on the average, but it consists of spikes 1000 times a second followed by no activity. Guess what? You will have now have a 1000 Hz jitter on your output! That's because it is the sudden change that causes the output waveform to change.
Of course, it is possible that even case 1 causes variations as there is no way to actually have a constant load. The system is running at billions of operations/second. How on earth do we characterize its load? Think: the timing counter on your player is causing CPU activity as often as it is updated.
So in that sense, PCs are disadvantaged here and one has a hard time putting much value on such assertions. It is best to side-step the problem altogether, not use the PC for timing at all using an synchronous, and isolated device which has a tiny microprocessor in it and not subject to such large variations and complexity as the PC.
Yeah, sorry for the cynical reaction, Vincent. I appreciate the approach as well, and given that the software is free, it even has some credibility (as opposed to the many jitter explanations from people selling very expensive jitter cures). And I don't expect the guy giving the software away to back up his theories/explanations with measurements and audibility. People like Sonic Studios and Empirical Audio should be doing that.
The oasis is still there, though. Jitter, even when it is audible, is a vanishingly small problem compared to the distortions that digital has vanquished. Digital is the great equalizer. And of course none of that keeps the witchengineers from running in to stir doubt in the minds of the obsessive compulsive.
Tim
Ok.
Save your wit for papers like this, I’m certain you will appreciate.
http://www.tnt-audio.com/accessories/bullet_vs_nextgen_e.html
Save your wit for papers like this, I’m certain you will appreciate.
http://www.tnt-audio.com/accessories/bullet_vs_nextgen_e.html
Ok.
Save your wit for papers like this, I’m certain you will appreciate.
http://www.tnt-audio.com/accessories/bullet_vs_nextgen_e.html
well at least he wrote that the differences are small far from the usual "Night and Day" or "blew away" ...
well at least he wrote that the differences are small far from the usual "Night and Day" or "blew away" ...
Baby steps...
I'm a little bit saddened to see this reaction, as Damien is one of the "good guys" who has made this available for free (although a paid version is now on the horizon). More importantly, the code is open-source.
There are several practical issues Audirvana addresses on OS X, which I will list in a somewhat subjective order (i.e., of importance to me).
1. The single biggest problem with iTunes is its inability to switch sampling frequencies on the fly. If you have Audio MIDI setup set to 44.1kHz, and use iTunes to play back red-book or most iTunes-store-provided music, it can do so with bit perfection. However, if it then meets, say, a 96kHz track in your playlist, it will downsample it to 44.1kHz on the fly during playback, unless you manually quit iTunes, manually reset Audio MIDI Setup to 96kHz, and then re-open iTunes. Audirvana (and several others, like Decibel, Ammara, Pure Music, etc, also will do this). This is primarily a technical advantage, but for those who take bit-perfect playback seriously, it avoids a potential degradation of sound quality.
2. It permits playback from memory. The entire file is decoded (and uncompressed) into memory, assuming sufficient memory is available. To enable seamless playback, the n+1 track is also cued into memory. Audirvana is not unique in being a memory player, but it has what is arguably the sanest implementation.
3. Like several others, it has the option of hogging the output device. This can be fairly critical, as Apple by default wants all applications to share the output device, so that you will be able to hear all of your mail arriving from Nigeria during playback of Beethoven's 9th. The resulting mixing, in addition to being an irritant, results in resampling the audio, again potentially degrading sound quality. Hog mode prevents this, and integer mode, available in 10.6 but not 10.7, in addition permits skipping two arithmetic operations, converting back and forth between integer and floating point, which are unneeded operations when one application is hogging the output device. I personally cannot hear any difference, but others claim to.
4. It can play flac, unlike iTunes.
5. It can up-sample more intelligently than core audio. Apparently, some DACs have a sampling frequency sweet spot, and this helps. I have never used this myself.
6. You can easily assemble temporary playlists on the fly (and you can also save them). (I seldom make playlists in iTunes because it is so cumbersome.)
There are a few others, but 1-4 and 6 are the most compelling to me, and having this available as an open-source project is an added attraction.
I've written a few Applescripts and an iTunes plug-in to enable using iTunes and its associated Remote.app with Audirvana, which are also available for free. (google nyquist.zsh ).
In any case, it is an application not without merit.
There are several practical issues Audirvana addresses on OS X, which I will list in a somewhat subjective order (i.e., of importance to me).
1. The single biggest problem with iTunes is its inability to switch sampling frequencies on the fly. If you have Audio MIDI setup set to 44.1kHz, and use iTunes to play back red-book or most iTunes-store-provided music, it can do so with bit perfection. However, if it then meets, say, a 96kHz track in your playlist, it will downsample it to 44.1kHz on the fly during playback, unless you manually quit iTunes, manually reset Audio MIDI Setup to 96kHz, and then re-open iTunes. Audirvana (and several others, like Decibel, Ammara, Pure Music, etc, also will do this). This is primarily a technical advantage, but for those who take bit-perfect playback seriously, it avoids a potential degradation of sound quality.
2. It permits playback from memory. The entire file is decoded (and uncompressed) into memory, assuming sufficient memory is available. To enable seamless playback, the n+1 track is also cued into memory. Audirvana is not unique in being a memory player, but it has what is arguably the sanest implementation.
3. Like several others, it has the option of hogging the output device. This can be fairly critical, as Apple by default wants all applications to share the output device, so that you will be able to hear all of your mail arriving from Nigeria during playback of Beethoven's 9th. The resulting mixing, in addition to being an irritant, results in resampling the audio, again potentially degrading sound quality. Hog mode prevents this, and integer mode, available in 10.6 but not 10.7, in addition permits skipping two arithmetic operations, converting back and forth between integer and floating point, which are unneeded operations when one application is hogging the output device. I personally cannot hear any difference, but others claim to.
4. It can play flac, unlike iTunes.
5. It can up-sample more intelligently than core audio. Apparently, some DACs have a sampling frequency sweet spot, and this helps. I have never used this myself.
6. You can easily assemble temporary playlists on the fly (and you can also save them). (I seldom make playlists in iTunes because it is so cumbersome.)
There are a few others, but 1-4 and 6 are the most compelling to me, and having this available as an open-source project is an added attraction.
I've written a few Applescripts and an iTunes plug-in to enable using iTunes and its associated Remote.app with Audirvana, which are also available for free. (google nyquist.zsh ).
In any case, it is an application not without merit.
Last edited:
Sorry for my contributions to the negative reaction, Scott. I'm sure it has merit and your first and last points are good ones. I don't have enough hi-res to care about native sampling rates, but the ability to just play a FLAC file without converting it first? That's nice. Anyway, sorry again for showing the dark side. Call me cynical, but this audiophile player thing, which is pretty dubious in the first place, all started out with Amarra, at $900, and reports on CA of how it had a dramatic, transformative effect on Mac playback (from people who were swearing their Mac playback was dramatically, tranformatively better than their CD player just hours earlier). And of course it was sound stage. It's always sound stage....well, when it's not micro dynamics. Just shoot me.
Tim
PS: I tried to try it. It downloaded but wouldn't install.
Tim
PS: I tried to try it. It downloaded but wouldn't install.
Well, I misread some stuff. It looked like everyone was piling on, but I re-read more carefully. My bad.
So you downloaded Audirvana? It is just a compressed archive (zip) of the application. I double-clicked it to unpack it, and double-clicked again to open the application.
I downloaded from this URL: http://audirvana.googlecode.com/files/Audirvana 0.9.5.zip
I am totally sympathetic to the cynical point of view, BTW. I am naturally predisposed, upon smelling flowers, to look around for the funeral. It does seem in this hobby everyone is trying to insert a syphon hose into my wallet.
I don't really hear much difference between the various playback programs. Audirvana seems a tad brighter in the midrange, but I doubt I could pick it out reliably in a double-blind test. Anything that makes a radical change to the sound would make me suspicious. I still cannot comprehend why identical files wouldn't sound identical using any bit-perfect-capable playback software.
So you downloaded Audirvana? It is just a compressed archive (zip) of the application. I double-clicked it to unpack it, and double-clicked again to open the application.
I downloaded from this URL: http://audirvana.googlecode.com/files/Audirvana 0.9.5.zip
I am totally sympathetic to the cynical point of view, BTW. I am naturally predisposed, upon smelling flowers, to look around for the funeral. It does seem in this hobby everyone is trying to insert a syphon hose into my wallet.
I don't really hear much difference between the various playback programs. Audirvana seems a tad brighter in the midrange, but I doubt I could pick it out reliably in a double-blind test. Anything that makes a radical change to the sound would make me suspicious. I still cannot comprehend why identical files wouldn't sound identical using any bit-perfect-capable playback software.
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |