Verify Bit Perfect Playback iTunes9 OS X 10.6?

[doubleij]

Can anyone verify that iTunes 9 outputs bit perfect audio on Snow Leopard? I tried to test with a DTS file, but iTunes doesn't have the codecs for DTS so I don't know one way or the other.

 

I've disabled Equalizer, Sound Enhance and Sound Check in iTunes. I've set the MIDI in utilities to 44/16 since that matches the quality of lossless audio in my library. My library is currently in FLAC format, but I am thinking of using XLD to convert to ALAC losslessly, managing and playing the library through iTunes.

 

I'm also willing to play the FLAC files directly through Cog or another player, if Cog or another player offers bit-perfect output.

 

Thanks in advance for your help. I'm a mac newbie.

 

[The Computer Audiophile]

"Can anyone verify that iTunes 9 outputs bit perfect audio on Snow Leopard?"

 

As far as I, and many others, can tell iTunes 9 on Snow Leopard is bit perfect. There are circumstances where one can make it un-perfect, but you have to try purposely to accomplish this.

 

[LIO]

With Snow Leopard iTunes uses the new library of QuicktimeX. To hear file DTS with iTunes you have to install Perian http://perian.org/ and install the file in the folder HD/System/Library/Quicktime! It's very easy... If you copy perian in the old Quicktime folder it doesn't works!

 

[bdiament]

Hi doubleij,

 

In my experience, "bit perfect" and FLAC are different subjects.

 

While expanding a FLAC in an off-line process can be shown to create a result that is a "bit perfect" match to the original, uncompressed AIF or WAV, performing the process in real-time, i.e. while listening, exacts a sonic price.

 

I recently ran some blind comparisons using some high res material, Redbook material as well as some original 24/192 and 24/96 recordings I made myself. Using encoders in different applications I have and trying different settings for the FLAC, I and some listening partners compared the results with the originals. In every case, it took only a few seconds for everyone to hear the subtle but easily audible hardening and brightening in the FLAC when compared directly against the uncompressed files.

 

Upon further reflection, I don't really find this surprising. In my experience, no non-linear coded audio format is going to sound as good as LPCM in direct real-time playback comparisons. Even with lossless, the audio data is yet another step removed from the audio clock.

 

What I do find surprising is that this isn't more widely noticed among the listeners who post in various forums and write for the various established journals.

 

So, I'm not sure of the value of "bit perfect" if the sonics are degraded. To me "bit perfect" means I'm getting the source without alteration. (Getting the bit values correctly but with timing errors -and I would not be surprised if this turns out to not be the only flaw- isn't "perfect".)

 

With hard drive prices as low as they are and dropping all the time, I've decided to build my own server using raw, uncompressed AIFs. What I've found so far (about 20% of my CDs loaded) is the sound is better - i.e. more like the master from which the CD is made- than it is when playing the CDs in any transport in my experience, regardless of price. That's what I want. Not a hardened version that sounds like an early CD player.

 

Just my perspective.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[cfmsp]

Barry, thanks for sharing...this is important information to some of us.

 

"I recently ran some blind comparisons using some high res material, Redbook material as well as some original 24/192 and 24/96 recordings I made myself. Using encoders in different applications I have and trying different settings for the FLAC, I and some listening partners compared the results with the originals. In every case, it took only a few seconds for everyone to hear the subtle but easily audible hardening and brightening in the FLAC when compared directly against the uncompressed files."

 

How obvious were the differences with just Redbook source material?

 

I don't have experience with FLAC encoders to know, as I've only used LAME's insane setting for my compressed files, but ... presumably FLAC encoders are all of similar 'quality' given that they are designed to decode back to the original bits (which is easily testable), and therefore your comparison results should be applicable across the universe of FLAC encoders?

 

 

 

thanks

clay

 

 

[pj]

I did some testing after reading a post on CA which postulated that iTunes and Amarra were using different filtering and this was what accounted for the "improved sound" of Amarra.

 

My test methodology was to route iTunes output direct to s/pdif out of a konnekt 24d with a cable looped back to the s/pdif in jack of the same interface. I played back a short 16/44.1 aiff file in iTunes and recorded the looped-back signal with Wave Editor.

 

- imported the original aiff in Wave Editor so both recording and original were 32bit files.

- added two channels to the original aiff and pasted in recorded data. File at this point contains four channels of data.

- used the first clear waveform peak to align the two files to sample accuracy

- selected both right channels and used WE subtract channels function to null the data. same for left channel.

 

The result was an empty data file with rms and peak levels of -infinty dB. This indicates that with the recommended settings (full volume, equaliser and crossfade etc switched off) that iTunes 9 under Snow Leopard is "bit-perfect".

 

Performing the same tests with Amarra demo enabled resulted in small (about 0.5 second) but audible artefacts at the start and end of the file but null data for the rest of the file. This may have been result of error on my part but methodology was exactly the same as with iTunes test. One possible explanation is that Amarra is performing a very short fade in and out on playback.

 

Based on the above it seems Amarra is outputting files that are identical (apart from as noted above) to the original .aiff which would mean that there is no difference in the output between the Amarra and iTunes in terms of "bit perfectness". Filtering of any type would alter the bit pattern and the two files would not null so is clearly not a factor in the claimed superiority of Amarra.

 

cheers

Paul

 

 

 

[bdiament]

Hi Clay,

 

"How obvious were the differences with just Redbook source material?"

 

On the Maggie system, the case was the same regardless of the original file's resolution. It took only seconds. In other words, as soon as the song started, the comments started. I would not call the differences "night and day" but they are sufficient that I wouldn't consider FLAC for my own personal use or for anything Soundkeeper releases.

 

Neither did it matter which app did the encoding or what the settings were.

For me, so-called "lossless" is out. Besides, hard drives are getting cheaper by the day. Only raw, uncompressed AIF (or WAV) for me.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[bdiament]

.

 

[cfmsp]

 

"In other words, as soon as the song started, the comments started. I would not call the differences "night and day" but they are sufficient that I wouldn't consider FLAC for my own personal use or for anything Soundkeeper releases."

 

thanks Barry,

this seems 'obvious' enough, although as you suggested drives are so cheap now (non-SSD anyway) this is a non-issue. I guess those storing music files on SSDs might think otherwise.

 

cheers,

clay

 

 

 

 

[Audio_ELF]

clay commented ... "this [Lossless vs Uncompressed] seems 'obvious' enough, although as you suggested drives are so cheap now (non-SSD anyway) this is a non-issue. I guess those storing music files on SSDs might think otherwise."

 

The question is ... is Lossless (FLAC or ALAC) stored on an (internal) SSD, better than Uncompressed (WAV or AIFF) on a spinning disk?

 

Eloise

 

[cfmsp]

 

thanks Eloise, a good question.

 

And please ignore my first reply here, it was a brain-fart, quickly erased.

 

One more point, fortunately for all of us, Barry used FLAC encoders, otherwise the FLAC aficionados would be able to say that Barry (& teams) observations only indicate that Apple Lossless does not sound as good as AIFF.

 

clay

 

[bdiament]

Hi Eloise,

 

For me there is no question. So-called "lossless" is at a loss from the start.

AIF or WAV from hard disk (the way many things are recorded nowadays) does not suffer (to my ears) the hardening and brightening.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[ggking7]

If Barry says it, I believe it. My whole collection is in FLAC too... wow. What about tagging? I don't think WAV files can be tagged. What about AIF?

 

Here's a thought though. If a DAC or sound card plays from its own buffer, why does the extra CPU time required to decode FLAC to WAV matter? It might make the data arrive a little slower to the buffer, but then it sits around for a while waiting for its turn to be played.

 

[cfmsp]

 

"If Barry says it, I believe it. My whole collection is in FLAC too... wow. What about tagging? I don't think WAV files can be tagged. What about AIF?"

 

AIFF supports tags.

 

If Barry says it, I believe it, too. But, if one cannot hear it for themselves, there's an interesting decision to be made. I have no doubt that Barry & team can hear things I might not be able to discern.

 

I tend to believe that my listening skills are average at best (for audiophiles). OTOH, I"m not averse to implementing a tweak I can't YET hear when the cost is negligible (to me).

 

I'm already converting my music over to AIFF, and haven't ripped anything to ALAC for quite some time.

 

YMMV,

clay

 

 

 

 

[bdiament]

Hi ggking7,

 

Thank you for the confidence but I can only report on what I hear. My best suggestion is to listen for yourself and draw your own conclusions. You may agree or you may hear it differently.

 

I'm not sure it is "the extra CPU time" that is making the difference. I tend to think it is the extra process, not in terms of load on the CPU though. Whatever the cause, my experience in the tests tells me it just isn't for me.

 

I've been using the AIF format as I load my server, the same format in which I record, edit and master. I don't know if AIF or WAV matters by the time you get the file into your server software. In my experience, some software seems to lose the tags, requiring retagging anyway. I chose AIF because I've been using it with good success for years.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[HiFiInsider]

Is iTunes bit perfect in Leopard?

 

[cfmsp]

 

It can be easily configured for bit-perfect, just turn off any sound enhancement features in iTunes (and check again with each subsequent iTunes upgrade), and ensure the sample rate in Audio Midi Setup matches that of your music files, and you're all set.

 

Were it not for the occasional changing of the defaults - for the sound enhancement features - in iTunes with each new release, it could be declared as bit-perfect out of the box.

 

cheers,

clay

 

 

[ggking7]

Barry, I'm a bit puzzled over your comment here:

 

"I tend to think it is the extra process, not in terms of load on the CPU though."

 

What difference can an extra process (or 10 extra processes) make if the digital data coming out is the same?

 

[bdiament]

Hi ggking7,

 

"What difference can an extra process (or 10 extra processes) make if the digital data coming out is the same?"

 

The fact is, the data is not the same - if you consider the timing component an important one, as I do. In other words, the "right" data at the wrong time is the wrong data.

 

The bottom line for me, is there is a sonic price to pay with non-linear files (i.e. so-called "lossless") where we get a step further from the clock that is so critical with digital audio.

 

Further, I'm not convinced this is the only thing going on.

Just my perspective.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[ggking7]

Hi Barry,

 

I don't see how timing can play into it considering that all sound cards and DACs actually output audio from their own internal buffer. From what i understand, instead of the data passing from the computer and straight through the DAC with no delay, it is actually stored in the DAC's internal buffer until it reaches the front of the queue and is then outputted. Whether it was originally stored as FLAC or WAV, the digital data sitting in the buffer should be identical in every way, and timing should be removed from the equation entirely as long as the buffer never runs out.

 

[bdiament]

Hi ggking7,

 

"...the digital data sitting in the buffer should be identical in every way, and timing should be removed from the equation entirely..."

 

In my experience, this is just another instance where (as I always say) "should" and "is" are two very different things.

 

If I haven't already said so (and I believe I have), I don't believe timing is the only issue.

 

Regardless, I and my listening partners hear appreciable sonic degradation when using a format like FLAC (again, this listening directly to the FLAC without first expanding it using an off-line process).

 

Have you made direct comparisons of your own?

What do you hear?

 

I don't say you shouldn't use the format if you like or that you or anyone else will hear the same things we do. (My experience says some will and some will not.) I'm simply saying, for whatever reason(s) I'm hearing what I'm hearing, data reduced formats aren't for me.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com