Pure Music

I just downloaded the demo version. It's changing the sample rate but not the bit depth. I'll wait until it outputs bit perfect audio before trying it again.

Hi,
Do you know if amarra does both?

Pure Vinyl and Amarra do both. I haven't checked this other program yet.

Poo 10.5 only

I really need to get off 10.4!!!

Anyone have a spare upgrade disc (PPC mac)

On the bit depth (word length), I find that neither Amarra Mini or Pure Music changes the bit depth. For example if I play a 24/96 track and then change to a 16/44.1 track, I see 24/44.1 in Audio Mini.

Chris, can you explain more about Pure Music not being bit perfect? How did you confirm this?

Hi bottlerocket - This is interesting. I'm not sure if I've found a bug or limitation in both Amarra and Pure Music.

When using the built-in optical output on a MacBook Pro or Mac Pro neither player can play a 16-bit song first followed by a 24-bit song and output 24-bits on the second song. I believe neither player can change the bit depth using the optical output. When using a Lynx card Amarra will report bit depth changes and it appears to be changing the bits. I'm testing this by sending HDCD encoded material to the Alpha DAC and checking the HDCD indicator. If the Least Significant Bit is not correct the indicator will not illuminate. The Lynx and Pure Music also appear to work bit perfect.

The Lynx bit depth is always set at 32-bit and there are no other options.

This raises questions about sending equal or greater than the max number of bits each track requires. If I set Audio Midi to 24-bits and use the optical output I can play 16-bit and 24-bit tracks and illuminate the HDCD indicator. Tacking on the extra zeros for 16-bit tracks is better than removing 8-bits from the 24-bit tracks.

I have run a ton of tests tonight. I'm still trying to process the results. I'm tempted to say neither app can adjust the bit depth and that's that. But, I think some of my tests indicated otherwise. Or, when I had the dCS Paganini DAC in here it displayed the bit depth of the tracks being played. I know for a fact it showed the correct rate for the tracks, but I think it just looks at the track and not how many bits are being sent by the audio card. If this is the case it would lead me to the lack of bit depth changing capability in both apps.

I hope other readers can run some tests.

Hi there Chris,

Or, when I had the dCS Paganini DAC in here it displayed the bit depth of the tracks being played. I know for a fact it showed the correct rate for the tracks,

I recall very well there was a great discussion about what the dCS showed, and it was the conclusion it could recognize the LSByte being all zeros, and thus shows 16 ("as siginifant" ?).

This may help too, at working it out :

I myself (XXHE) spit out 24 bits (or 32 actually when the device allows it), no matter the source material is 16. This is because you *may* want to use the volume, and then it starts using those other 8 bits. But if not, the least significant byte remains zero, and all HDCD receivers I have heard of, show "bit perfect", also those who say to receive 24 bits.

Now, we may even wonder what it means - or what we want when two subsequent tracks change of bit depth; In the case of XXHE the audio engine is restarted and physically the device is addressed differently. BUT, what does not change (and cannot change I think) is that you, the user, in between state that your device wants to receive 16 bits only, once the before track was 24. Because (this may be hard to get) a 16 bit track always has the potential to "need" 24, and this is again when the volume is needed (or other DSP stuff has to jump in on user command). So, in advance you state what the DAC should receive (16 bits or 24 bits), and the playback is going to behave like that. This is unrelated to the source material ...

The above is how I do it, and I don't see it go otherwise while staying praticle.
The most important might be that either case is "bit perfect", as long as there's two bytes only, or when more, the LSBytes are zero (for 16 bit material).

Of course, when a first track is 16bits and a second is 24, while the receiver "stays" on 16, things are very wrong.

I hope I made clear what I wanted to make clear !
:-)
Peter

Thanks for that reply Chris. I resolved to 1 time set my MAC to 24 bit using Audio Midi and then Amarra Mini just switches the sample rate back and forth between 44.1khz and 88 and 96 etc. I think this matches what Peter suggests too. Makes sense to me that its better to zero fill a 24 bit word when going from 24 to 16 bit, than to truncate at 16 bits when going from 16 to 24.

BTW, I am using a MAC digital optical out.

I asked this question to Amarra if they switched both the word length and the sample rate or just the sample rate and they never answered me.

thanks Lars,

Do you know how to compare Pure Music to iTunes (a la Amarra)?

once Pure Music is opened, I can't figure out how to play music via iTunes only - e.g. to switch back and forth from one to the other.

More importantly, when Pure Music is closed, I cannot play music via iTunes without restarting it first. Pure Music does not seem to reconnect the output stream to iTunes upon it's own shutdown.

The combination of these two issues makes it impossible to instantly compare one app to the other. The closest I can get to that feature of Amarra is to start a song in iTunes, then open Pure Music, which re-starts whatever song was playing at the beginning.

There's also a bug whereby Pure Music does not recognize the new song selected in iTunes, and continues playing the original selection, even though iTunes indicates that it's playing the different selection. This continues even after stopping and re-starting the song via iTunes.

Any experience with either of these issues?

it sounds different in my system than the earlier version of PV I tried.

thanks
clay

One thing I noticed with PureMusic is that with the Hiface it is crackle city, whereas PV is maybe an occasional pop, but generally good.

Hi Clay,

I have not listened to Pure Music and probably won't try it. Pure Vinyl is working very well for me, but I haven't tried the latest version just posted. Anyway, I'm getting a bit worn out testing all of this stuff. It's time to enjoy some tunes.

Steve

enjoy yourself then, Lars :)

I'm rather tired of trying to 'hear' differences with all the changes I've incorporated as well.

clay

Well I for one have been really happy w/ pure music today...

Was comparing pure music side by side w/ amarra (just load some songs into amarra's playlist mode and play them, then play the same ones with pure music). Feeling like amarra definitely has its own certain sound - smooth, but condensed. Sometimes this is better. And the equalizer is important. Pure music takes away the harshness of itunes but allows the spacing of the instruments and what feels like more interesting and vinyl-like timbre. Could it just be that this new sound has caught my ear b/c it's different from what I've been listening to and old songs sound new again? Will have to see how this holds up over the next few weeks...

So, it appears that the Pure Music Version does not have the Upsampling feature of Pure Vinyl. I know that many audiophiles don't want that feature, but some people may want it and it does not appear to be there. Just thought I would point that out.

My Bad... It appears it will be in the final product.

"Note: this option will be disabled until the final (non-Beta) release of Pure Music, which will
include the same high quality 64 bit real-time upsampling feature currently included in the iTunes
Music Server feature of Pure Vinyl."

Just an small input from my side. I am back home, so I had a chance to test the new Pure Music 1.0b1, on Mac OS X 10.6.2.

It us 100 % Bit True for 16 and for 24 Bit, for AIFF and WAV (I haven't tested ALAC), it does automatic Sample Rate change (tested 44.1, 48, 88.2 and 96K), forward and backward, with and without memory play, and is Bit True in all tested sample rates.

Juergen

Hi Juergen - Did Pure Music switch he bit depth for you? If you use an audio device that can be set to 16-bit I don't think PM can change it to 24-bit when needed (Amarra is the same way). Maybe there should be something in the manual that say set Audio Midi to 24-bit.

Chris: I am not sure, whether I understand your question correctly, but I set the Bit Depth in the Audio MIDI Setup to the resolution of my Audio Interface or DAC.

In case I have a 16 Bit Interface or DAC, then I will test only 16 Bit Signals, because 24 Bit signals will be truncated to 16 Bit, so if then 16 Bit is OK, everything is fine.

In case this would be a 24 Bit Interface or DAC, then I will test 16 Bit and 24 Bit Signals for Bit True. I will not look back at this Audio MIDI Setup, I just set it to 24 Bit and if 16 Bit Signals came through with 100 % Bit True, this means that Bit 17 to 24 is stable Zero all the time, so it would make no difference if I would have set the Audio MIDI Setup to 16 or to 24 Bit.

Is this what you where asking? Because why should a software switch bit depth, because the software didn’t know the resolution of your interface of DAC, so you should set this correctly. This is the same with WASAPI under Windows 7 or Vista. So in WASAPI you have to set the same bit width, as your connected interface or DAC.

Juergen

Hi Jeurgen - This comes into play with an interface like the built-in optical output on a Mac. It supports 16, 20, and 24 bit output. I've always thought it best to output the same bit depth as the music being played. Thus thought Audio Midi should switch between 16 and 24 bit to be exact. My Lynx card only outputs 32 bit so there are no options.

Isn't this a combination of what the DAC and the digital I/O support, not just what the DAC supports? Maybe I'm making a bigger deal about this than needs to be made and everyone should just set the bit depth at 24-bit if possible.

I do not think that there is any downside to outputting 24 bits from the player to the DAC for 16 bit files, as I believe 16 bit files are just padded with zeros, and the native 16 bits are not changed in any way. In fact, some engineers have mentioned to me that many DAC chips will sound better getting the 24 bits even with 16 bit source material (not sure I understand why though, something to do with noise shaping maybe, perhaps a true expert like Gordon Rankin could answer this question...) I think one will find that when outputting 24 bits from a 16 bit file, the 16 bits stay exactly the same (bit perfect) with 8 zeros added below for properly designed player software.

If the software is working correct, than this setting is absolutely signal independent (as what PeterSt was also saying), because then everything above the signal bit width is filled with static Zeros.

You have to set it to the minimum value of your signal chain that is behind your playback software, so the interface and the DAC. A BiPhase signal is limited to maximal 24 Bit, so this is the case with SPDIF, AESEBU or TOSLINK, so here 24 Bit would be correct.

Even if the Lynx Card supports 32 Bit between the software and the card, at the digital out of the Lynx Card to the Berkley DAC (for example) you can only transfer max 24 Bit, so also here 24 Bit would be correct, if you want to stay Bit True.

And also in the case you use a digital volume control between the software and the lynx card, you should and could use 32 Bit, but when you finally want to transfer the digital signal to the Berkley DAC, you need to set the dither of the digital volume to 24 Bit.

This has nothing to do with the internal bit depth of software for EQ or Volume Control, because this resolution is prior the point, where the software gives the signal to the outer world.

Does this answer the question?

Juergen

Borrows: When connection the digital devices via BiPhase (SPDIF, AESEBU or Toslink), than it does make no difference, whether you are transferring for example only the 16 Bit signal or the 16 Bit signal with padded 8 Zeros.

But internally with I2S interfaces (or in some very rate external I2S interfaces), there is a difference, whether you only transfer the 16 Bit signal, or the 16 Bit signal with padded 8 zeros. But to go in this field deeper, it will be too complicated, because then you have to split up again, between very rare NOS DACs and the typical oversampling DACs, or digital oversampling filter. So I will stop here.

Juergen

Slightly off topic, but I was wondering what people consider to be the optimal interface for the Berkeley Alpha DAC? I use a Lynx AES16, but know from trying an Antelope DA that the Lynx produces sufficient levels of jitter and noise to audibly degrade the Berkeley's performance. I see some use the Legato (async USB/SPDIF) and others use an external clock for their Lynx card. Berkeley is working on their own AES reclocker so they agree the Lynx on its own leaves room for improvement.

Would appreciate the benefit of others experimentations. Thanks.

David

Earflappin: The lynx is a very good sound card with automatic sample rate switching, but you are right and I can confirm, concerning jitter, leaves room for some improvement. When using external clocks, this improves the jitter and sound, but then no longer automatic sample rate change.

Juergen

Juergen, thanks for the response. For as many people who use the Berkeley I'm surprised to see so little discussion regarding the most optimal interface either here on this forum or on others.

Anyway, I'm going to do some experimentation. I know the Antelope DA provides improvements, but it does not reclock the AES. I'm going to get a Black Lion Micro Clock Mk2 to try. Socrates is coming over next week and will bring is Legato so I'll get a chance to hear an async USB/SPDIF interface. Then when Antelope comes out with their Zodiac Plus next month I'll get to hear what a reclocked and noise isolated AES interface sounds like. I'll report back my findings.

My iPod Touch timing bar does not move when controlling PM. Anyone else notice this?

Rob explains that iTunes is not really being used (as he reminds us it is in Amarra, a waste of CPU in his mind). PV or PM can make iTunes timebar work on the main GUI (set in preferences) but the remote app must look at a different aspect of iTunes and display a static time remaining bar in the iPod/touch (the bar changes to last seen time if song is paused for any reason, but remains static there again). Also, as documented, the play/pause indicator does not change (it works, just doesn't change), again for the same reasons...iTunes is not playing the song...PM/PV is.

I have been told by an acquaintance that Pure Music will play at the word length of the source IF its volume control is left at 0.

Can anyone verify this?
(I have not yet downloaded it to demo.)

Thanks in advance.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

When using the volume control in iTunes, the volume setting is mapped to a
corresponding volume setting in Pure Music (the iTunes maximum setting is
mapped to 0 dB in Pure Music). iTunes is not controlling the volume, because it
isn't playing the file. With the Monitor setting in Pure Music at 0 dB, the
original audio isn't altered at all: you get direct playthrough via a lossless, 64
bit audio kernel to your high-end DAC.

Barry, see my comment above.

PM is bit true for 16 and for 24 Bit signals, so when you put in a 16 bit signal, then absolutely the same 16 bit signal gets out.

Juergen

I raised the question because of Chris' comment early in this thread, where he suggested that sample rate was changing but word length was not. (I'm not sure how this was tested. I don't believe the HDCD indicator on the Alpha is an accurate way of assessing this -- if indeed that was how the determination was made... I don't know. I prefer an accurate bit meter like the one in SpectraFoo.)

To clarify, if I switch between a 16/44 file and a 24/192 file, I want both to play back natively. And I shouldn't have to mess with Audio/MIDI settings.

I don't want to pad the 16-bit file with zeros, don't want the 24-bit file truncated and I want the original sampling rate.

So, the question is, can Pure Music do this?

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Here's another question for users of Pure Music:

If I understand correctly, the app uses iTunes interface and playlists but iTunes is not running (as it is with another app that uses iTunes as a "back end").

If this is indeed the case, if a tune is selected in the iTunes window and the user accidentally hits Delete, can the tune be deleted?

Can *any* changes be made to the iTunes library when Pure Music is running?

Just curious. When my library is fully loaded, I hope to write protect the disk that houses it and so avoid any accidental changes or deletions.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Barry, everything goes out as it goes it. So absolutely the same signal at the output as on the input. (PS: padding with static zeros = zero). Nothing added, nothing lost.

Juergen

Hi Jeurgen,

I understand padding with zeros is benign but if the source is 16-bits, I do not consider a padded 24-bit file the same as what goes in.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Barry,

Why not give Rob a call at Channel D or e-mail him. I'm sure he will be happy to answer your questions.

Hi Lars,

Thanks. Good idea.
Done. I await the response to the questions I sent.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

I received this response today from ChannlD:

"Sorry for the delay in responding, generally we respond the same day, if not within a few hours, but we are short staffed because of exhibiting at the AXPONA expo last week, and things are just now getting back to the normal state."

In response to my issue that simply quitting Pure Music prevents any iTunes output without a restart:

"Have you tried slightly nudging the volume control in iTunes after quitting Pure Music, and then playing iTunes?" So, others with this issue may want to try this recommendation.

In response to my issues of loss of itunes-PM synchronization & intermittent stopping of play for no apparent reason:

"We are getting reports of occasional synchronization issues similar to those that you have reported. This happens especially when jumping around frequently in a playlist. We are working to correct this problem and appreciate your patience. In our internal testing, if allowed to play through a music library on its own, the test systems have been stepping through tracks continuously for at least several days. Can you be specific about the track format (sample rate, file format), if the tracks are gapless, and if you're using Memory Play (and Hybrid Mode, etc.)?"

While I've not had a chance yet to try the suggestions, I wanted to post what was a welcome reply to me.

Our patience will likely result in an improved experience with Pure Music, and if Lars' experience is repeated, we may not have to wait long.

Clay

Couple of comments.

Even when deleting files from iTunes, one has to respond actively to an additional prompt to move the file to Trash. The standard delete in iTunes simply deletes the file from the iTunes library, but does not delete the file. Of course, even after the file has been moved to the Trash, it has still not been actually deleted from the disk.

Regarding the integration with iTunes - Channld has chosen a less tight level of integration with iTunes than did Sonic with Amarra. Unlike Amarra which attempts to 'integrate' with all iTunes functions, Pure Music only link to iTunes for a limited set of functions. This allows ChannlD less onerous access to the iTunes library / file management, without the headaches of full-on integration.

Personally, I never understood why Amarra's integration required the simultaneous playing of songs via iTunes. Now that (I think) I better understand why, and the resulting tradeoffs, I feel that the ChannelD approach is better, given the rather larger processing footprint required by Amarra/iTunes. Of course, at the time Sonic made their decision, it was not as widely accepted that a significant portion of the differences we hear between different computer sources is due to the processing footprint of the music player itself.

YMMV, as always,
Clay

PS, there are at least two other music players being built - one by Audiofile Engineering, and another of unknown source, but referenced recently on AA by Gordon Rankin. Apparently this new player will eschew any floating point conversions, and will bypass CoreAudio to a greater extent than other programs in existence.

Hi Clay,

Yes, I'm aware of Twilight. Don't know the other one.
Why would anyone perform "floating point conversions" in a player anyway? (I can't think of a good reason.)

As to deleting from iTunes, I'll likely just write protect the (separate) hard disk that houses the iTunes library, once I have it all loaded. I hate to think any of the work involved in loading the library would have to be done again, so I want to keep it safe from accidents. (I also have another disk that serves as backup, just in case.)

Meanwhile, I'm looking forward to trying out Pure Music this week. The advance word from some colleagues have been very enthusiastic. I'll run it through the Maggie system and see what they tell me.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

So far, Pure Vinyl is the best I've heard of any player. I will install Pure Music this weekend and compare the two for everyone. I suspect that they will sound quite similar.

Gordon stated:

"The only real reason to use Float32 is for mixing with other applications which gives any Audio Stack and easy way to do that."

Amarra, Pure Vinyl (Pure Music), and Twilight are all derived from music editing programs. This may explain the use of floating point conversions.

Steve

"The only real reason to use Float32 is for mixing with other applications which gives any Audio Stack and easy way to do that."

I think the more relevant reason for our music playback needs is to support dithered volume control.

clay

"Personally, I never understood why Amarra's ............., given the rather larger processing footprint required by Amarra/iTunes."

Try checking your MAC activity monitor comparing Amarra Mini and Itunes to PM and Itunes. Both run Itunes, but PM only uses the database and not the player. However PM uses way more CPU than Amarra - so much so that the combination of Amarra and Itunes playing simultaneously takes less CPU than PM and Itunes. Try it. I am not buying Channel D's "footprint" claim. Once you try the above repeat it with PM running playing from memory. I hit 48% on CPU doing that. Amarra and Itunes together was about 8%.

I like PM though. It's a good improvement over PV. Simple and the nasty clicking I was getting with PV is gone. The price is right too.

Running hi-rez files in memory with PV I see about 4.5% CPU using a 2.93GHz Core 2 Duo. It is less for Amarra. I have never heard clicking on any PV version. Perhaps this may be related to your computer's resources.

"However PM uses way more CPU than Amarra"

I remember reading - probably release notes/instructions for PM - a list of features one could turn off to reduce unnecessary (for music playback) processing by Pure Music. As I recall, it was prioritized by the amount of processor reduction. Employing these recommendations might dramatically change your experience of PM.

One thing I never liked about PV was the gratuitous (in my opinion) video display, e.g. the spinning vinyl metaphor & the firefly whatever. Thankfully, PM has somewhat less, but even that should be turned off if processing is a concern, and especially so, if you're running headless, as the VNC server also creates quite an unnecessary extra load.

YMMV,
Clay

Good points Clay. Also, maybe the upsampling was on to hit such a high CPU.

I believe upsampling is scheduled for a future release of Pure Music.

getting back to the iTunes integration approaches (of Amarra vs. PV/PM) for a moment, I believe that the downside of ChannlD's approach is that the control synchronization (between iTunes and Pure Music) will likely remain its achilles heel. IOW, it's likely that - with Pure Vinyl/Pure Music - those (like myself) expecting to be-bop around through their iTunes library will be more likely to lose sync between the iTunes and PV/PM apps.

ChannlD has this to say about this topic:

"Note to audiophiles about the iTunes Music Server Feature:

" [Pure Vinyl/Pure Music] provides you with direct access to an ultra high quality music playback engine. However, as with computer software, there is a very, very strong temptation to "test" it by "exercising" the controls, skipping forward, back, etc. repeatedly, trying to provoke hiccups. That is, in essence, using it as a toy, instead of as a tool for serious music listening. (Since critical listening is impossible while rapidly stepping through tracks, etc., that activity is best left to native playback with iTunes, etc., anyway.)

When using [Pure Music], envision how you would use a standalone, hardware based CD player (especially one having a memory play feature, which has the same side effect of delaying playback while the CD is loaded into memory). Rather than constantly exercising the remote, just press Play, relax, and experience the superb, crystalline, highest possible quality playback available for your music..."

While I certainly appreciate the intended 'spirit' of the Channel D comments above, I also happen not to agree with it completely. One of the joys (for me) of computer access to our music libraries is instantaneous access to each and every track. Channel D's "philosophy" assumes that listeners are willing to remain subject to the limitations of the previous generations of technology, i.e. CD / Vinyl playback.

In some ways, it's a great example of what we in the software development industry refer to as "calling a bug a feature". It's certainly a preemptive strike against those who might take issue with the rather, shall we say, sensitive nature of the user interface, esp. with respect to maintaining control synchronization with iTunes.

I can certainly respect their approach, but I'm hopeful that with Pure Music we will see improvements in this area, as I, for one, don't expect to give up the newly found freedom of easy, instant access to all of my music tracks.

As Channel D also says at the end of their notes to audiophiles,
Enjoy,
Clay

I have have no problems getting access to my library with Pure Vinyl using Remote with the iPodTouch.

Clay,
I totally agree! I love the sound of PV, but one of the main reasons I got into computer audio was so I could use the shuffle feature which I grew to love using my Ipod. Using shuffle with PV is very frustrating as it loses sync very often and PM works no better! If another reasonably priced alternative comes along that sounds as good and doesn't have this issue, I would quickly change to that! Hopefully PV will finally resolve this issue!! I'm sure with the excellent pricing of PM, many should sell.

Barry,

I understand padding with zeros is benign but if the source is 16-bits, I do not consider a padded 24-bit file the same as what goes in.

Juergen is correct in saying that padded 16 to 24 bits is not an issue and zero=zero. A better way to say it is the end analog output would not be any different for 16 bit data padded with zero into 24 bit data or outputting the same signal in it's 16 bit form.

Chris, yes you should set the Audio Midi to 24 bits for best results.

I don't really know what Rob is doing here but a little insight into Core Audio and such is that most applications use Float32 internally. There is a move to go to Float64 as there is no real way to carry 32 bit integer PCM data bit perfect with Float32.

So in most cases a sample at 16 bits is converted to Float32 if nothing touches it then Core Audio will look at the output device see if it needs to resample (hopefully not) and what the bit size is and then convert the Float32 to an integer. When converting a 16 bit sample to an integer 24 it will in a sense pad the sample with zero.

I like PM and I do agree it sounds better than Amarra especially with the Memory Mode turned on. I will say that iTunes is still running and still eating up processor and huge amount of virtual and real memory. Running top in the terminal I still find that with Amarra and PM that iTunes takes up about the same resources and processor time. But PM over Amarra is a lot less stressful on the Processor by between 2-6% depending on the speed and resources of the MAC.

I did receive another Twilight beta last week. I will load that up and see where that is at in my free (yea right) time.

Thanks
Gordon

Shuffle is not allowed at my house. :-)