Jump to content
  • The Computer Audiophile
    The Computer Audiophile

    Apple Music's Lossless and Hi-Res Mess

     


    This morning I've been testing Apple Music's new lossless and Hi-Res offerings on both my iPhone 12 Pro running iOS 14.6 and my Mac Mini (M1) running macOS 11.4. In my tests, I'm only concerned with playing the music bit perfectly, in other words without making any changes to the audio. If Apple Music says it's streaming lossless audio, then I want to stream that audio losslessly, rather than accidentally converting it to lossy AAC or MP3 etc... Whether or not people can hear the difference is a topic for another discussion. I'm just making sure I can play the music in its original form and that Apple is sending true lossless and Hi-Res to my audio devices. 

     


    Let's Dig in


    What is bit perfect and why should I care? In the simplest terms, bit perfect means that the audio hasn't been changed. The music sent, in this case from Apple Music, into the playback device hasn't been altered. The source is what has been delivered to Apple by the record labels. Apple is just the delivery company.

     

    If you care about high quality, getting the lossless streaming you're paying for from Apple Music, Qobuz, Tidal, Amazon Music, etc... then you should care about bit perfect because without it you have no idea what's happening to your audio. If this isn't a concern for you, no worries. 

     


    Testing Methodology 

     

    Device 1
    My Apple iPhone 12 Pro running iOS 14.6 and the Apple Music app. I connected the newest version of the Apple Camera Connection Kit to the iPhone, so I could attach a USB Audi interface, and feed power to the phone and interface. 

     

    Device 2
    Apple Mac Mini (M1) running macOS 11.4, and Apple Pro Display XDR, and USB audio interface connected to the ports on the back of the display. 

     
    I use the following testing methodology to test Apple Music. 

     

    The USB audio interface is a Berkeley Audio Design Alpha USB that accepts USB input and outputs audio over AES/EBU or S/PDIF (BNC). 
     
    I use a Berkeley Audio Design Alpha DAC Reference Series 3 that identifies and decodes HDCD on all sample rates from 44.1 up through 192 kHz. When an unaltered HDCD music track is played, the HDCD indicator on the DAC is illuminated. The HDCD flag is on the 16th bit for lossless CD files and the 24th bit for high resolution files. Any alteration, DSP, volume leveling, etc... changes this least significant bit and won't enable the HDCD indicator to illuminate on my DAC. Apple Music's lossless audio that I tested was 16 bit / 44.1 kHa and the Hi-Res audio was both 24 bit / 176.4 kHz and 24 bit / 192 kHz. That's the hardware piece. 
     
    With respect to source files, here's what I do. 
     
    I have a list of roughly ten known HDCD albums (although I could use more if needed). Many of these albums were only released as HDCD encoded CDs/files. There is no alternate lossless version. For example, Reference Recordings only releases CDs that are HDCD encoded. Pearl Jam's Live On Two Legs was only released as an HDCD master for its lossless CD version. 
     
    I set a baseline by playing my own local copy of the albums and make sure the HDCD indicator illuminates. I played some Reference Recordings albums through Apple Music on macOS and made sure the app could handle bit perfect playback. All was good there. On iOS, I used used other apps such as Qobuz, to play the identical music through the identical hardware. All was good through the Qobuz app. Again, there are no alternative versions of these lossless albums. It's the same music on all the services that offer lossless streaming.
     
    Absolutely there are possible holes in my methodology, but I believe I've minimized them as much as possible. If anyone can identify a hole, please let me know and I will retest. 
     

     

    Test Results
     
    On macOS, I found no way to play bit perfect lossless or Hi-Res audio from Apple Music. In addition to a couple other nonsensical issues that I'll get into later, Apple is doing something to the music it streams. 

     

    Test 1, streaming Pearl Jam's Live On Two Legs release I was able to illuminate the HDCD indicator for the first couple seconds of playback. After this, the light went out for good, even if I skipped to the next track. When I clicked the play button to start the entire album over again, the HDCD indicator illuminated again for a few seconds. If I had to guess, I'd say this is because of watermarking mandated by the major record labels. Apple has a perfect copy of the album on its servers, the perfect copy starts, but then something changes in the stream that causes the music to not be bit perfect. I'm open to all input on what this could possibly be, but watermarking is my best educated guess for now. 

     

    Test 2, streaming the Reference Recording's album Exactly Like This from Doug MacLeod, displays different behavior and bolsters my aforementioned watermarking theory. This album, from a very small independent record label that I don't believe watermarks it's music, alters between bit perfect and not bit perfect. Upon playback, the HDCD indicator is on sometimes then off for a period of time, then back on etc... I really don't have a good guess for why this happens. I originally thought maybe an adaptive bit rate issue caused it, but even after downloading the tracks to my device offline and playing them, the problem remained. 

     

    Test 3, streaming the Reference Recording's album Break The Chain from Doug MacLeod produced the identical behavior. The only difference here was that the album was Apple's Hi-Res offering at 24 bit 176.4 kHz. Again, no solid bit perfect playback. Something is happening to the music. 

     

    A note about Apple's Hi-Res offerings that makes this a real mess. On Macs running macOS / OS X, the Apple Music app looks at the sample rate in Audio Midi upon the app's launch. Whatever same rate is set there, is the sample rate that Apple Music will use for playback as long as the app is open. OK, fine you say, Apple Music lossless is probably 16 bit / 44.1 kHz and that can be set in Audio Midi. Sure, now for the mess. Apple Music Hi-Res is be definition not 16 bit / 44.1 kHz. It go up through 24 bit / 192 kHz. OK, you can run through the whole dance closing Apple Music, manually setting the sampan rate in Audio Midi, then reopening Apple Music and off you go. Oh no you can't. Apple Music doesn't tell you what the sample rate of its Hi-Res music albums. Thus, you have no idea at what sample rate to set Audio Midi. 
     

    UPDATE: To find the sample rate of the album and play it correctly you have to play a track, click the info button to reveal the sample rate, change Audio MIDI to correct sample rate, restart Apple Music, play the track again. 

     

    How did I find the same rate? Fortunately, the Apple Music app on iOS has auto sample rate switching, which enabled me to get the rate, then sixth back to my Mac to run the tests. This was good because I could see the bit perfect audio stream for the first few seconds once I had the correct sample rate set. 


    Note: the Qobuz app plays this music bit perfect on my Mac.

     

     

    On iOS, as I just mentioned, we have the high benefit of automatic sample rate switching when playing music in Apple Music. The results for Test 1, 2, and 3 were identical on iOS as they were on macOS. Bit perfect for the first few seconds of major label albums. Bit perfect on and off for Reference recordings' albums. Apple Music on iOS switched between outputting 44.1 to outputting 176.4 without an issue. If only the audio would remain bit perfect during playback, it would be a great solution. 

     

    I will note that the Qobuz iOS app played everything bit perfect, but there needs to be an asterisk. For some reason Qobuz resamples the 176.4 Doug MacLeod album Break The Chain at 192 kHz on iOS rather than 176.4. I checked Neil Young's greatest hits to make sure I cold stream 192 material bit perfectly from Qobuz and succeeded. iOS and iPhones are fully capable of bit perfect audio at 176.4 kHz, so I'm not sure why Qobuz is resampling the RR releases. 

     


    What About mQa?

     

    There is interesting news on this front. Some labels have snuck mQa material into Apple Music just like they have on other services. Users of Apple Music can search for mQa and they'll see some albums such as the Radka Toneff Fairytales album. The albums playback as mQa on a DAC in my system that is a full mQa decoder. 

     

    I checked a number of other albums that have appeared on Tidal as mQa only and didn't find mQa for these releases on Apple Music. 

     

    As readers of Audiophile Style know, I'm no fan of mQa and am very pleased it hasn't made its way into more releases or officially into Apple Music. Those who may be reading this as fans of Apple Music, rather than typical audiophile offerings, and aren't familiar with mQa, can get the gist of it and the company from the following videos.


    Part 1 - https://youtu.be/pRjsu9-Vznc
    Part 2 - https://youtu.be/NHkqWZ9jzA0

     


    Wrap Up

     

    As it stands now, Apple Music's lossless and Hi-Res offerings are a bit of a soup sandwich. You can't really stream the audio without some type of DSP going on that makes the music different from the lossless version on CD and on other lossless services. My guess is digital watermarking. In addition, it's not possible to get a consistent lossless or Hi-Res stream for other music that I tested, such as that from Reference Recordings. When I ran into similar issues with Amazon Music HD, using its apps just like I used the Apple Music apps, I was happy to find the Amazon Music HD streams lossless and Hi-Res through third party devices from Bluesound. Given that Apple doesn't integrate with Bluesound, I can't test this. Apple does integrate with Sonos, but as I found previously (link), the new Sonos Port can't stream bit perfect either, so a test on that platform would be useless. 

     

    I will happily update this article if there are holes in my tests or something else changes. As it stands now, I don't know of any holes and I stand by these conclusions. 

     

     

     




    User Feedback

    Recommended Comments



    10 hours ago, jkevin said:

    Just curious, in all this testing has anyone actually listened to any music?

    Yes. I'll do an half hour of each AM and Qobuz or switch between them at low or high volumes. Qobuz always has a tighter image and deeper stricter bass. AM has a wash to the treble that you only notice if you switch to Qobuz of the same track.

    Share this comment


    Link to comment
    Share on other sites

    2 hours ago, pkane2001 said:

    The result?  Apple Music stream is 0.2dB lower in level than the original track I own, with an apparent application of dither after the level reduction. This accounts for why they are not a bit-perfect match. There is no other filtering or perceptual coding applied. The files are an extremely close match when reduced to about 23 bits.

    Interesting. I believe this to be the case. The dither reduces the image as it is a random stereo dither. It also degrades depth because it's quantizing dynamics, like resampling it's never a free lunch.

    Share this comment


    Link to comment
    Share on other sites

    8 hours ago, pkane2001 said:

    Out of curiosity I tried a recording of a 24/96k Apple lossless Hi-Res track that I already owned in this format.

     

    Used iPad to stream through the Music app, and an all digital recording into an iMac.

     

    Out of curiosity: how did you manage the recording session between the iPad and the Mac?

    Share this comment


    Link to comment
    Share on other sites

    @wklie I listed Loopback app in my previous reply when you mentioned a virtual soundcard driver.

     

    FYI, I have Loopback installed on my machine. Keep in mind that a slight latency is involved: ~20 ms.

     

    P.S.: did you take a look at my Roon message covering the iPad recording? It's a too complex process, isn't it?

    Share this comment


    Link to comment
    Share on other sites

    When I mentioned virtual soundcard driver I assumed it will somehow allow directly capturing of the bits (i.e. performs like stdout from shairport-sync), but I don't know whether this makes sense or not as I have not looked into this, and I'm not familiar with Mac OS.

    Share this comment


    Link to comment
    Share on other sites

    On 6/16/2021 at 10:29 AM, wklie said:

    By streaming Fairytales (MQA) from iPadOS 14.6 Apple Music → AirPlay1 Lumin (at max volume) USB output → Full MQA DAC shows MQA. (successfully authenticated MQA, not possible with lossy encoding)

     

    As we know by now, AirPlay1 of iOS Apple Music 16/44.1 is bit perfect as I discovered above and confirmed by others.

     

    Not that this is adding new findings, just for the sake of completeness of testing MQA: I have changed the above test from AirPlay1 to iPad USB output via Apple Lightning to USB3 CCK.  Same iPad unit.  Same iOS.  Same track.  Same Apple Music settings.  Same USB MQA DAC.

     

    MQA light never shows up, so the integrity of 16/44.1 lossless is destroyed via iPadOS 14.6 USB output.

     

    As a control to prove the CCK and the iOS itself, Tidal app on the same setup successfully shows MQA on the external USB DAC (after disabling loudness normalization, and optimized playback which are on by default.  Crossfade is off).

    Share this comment


    Link to comment
    Share on other sites

    1 hour ago, wklie said:

    When I mentioned virtual soundcard driver I assumed it will somehow allow directly capturing of the bits (i.e. performs like stdout from shairport-sync), but I don't know whether this makes sense or not as I have not looked into this, and I'm not familiar with Mac OS.

     

    It's interesting to observe that the Shairport Sync recording via stdout compared to the original FLAC file in a multitrack session (all three 16 bit) isn't bit perfect.

     

    The result is audibly muted, the volume levels the same and yet the Frequency Analysis shows that it's not perfectly nulled.

    Share this comment


    Link to comment
    Share on other sites

    2 hours ago, Marco Klobas said:

     

    Out of curiosity: how did you manage the recording session between the iPad and the Mac?

     

    I used an iPad camera adapter to USB input on the DAC that has a Toslink digital out (RME). My Mac is an old one that has Toslink input built into the microphone jack, so that was easy to record from the DAC output. The other way I could've done it is use another interface that has digital input and firewire output to Mac (Apogee Element).

     

    Share this comment


    Link to comment
    Share on other sites

    2 minutes ago, pkane2001 said:

    My Mac is an old one that has Toslink input built into the microphone jack, so that was easy to record from the DAC output.

     

    Clever.

     

    My relatively recent Mac (2018) doesn't have TOSLINK (neither in nor out).

    Share this comment


    Link to comment
    Share on other sites

    16 minutes ago, Marco Klobas said:

     

    Clever.

     

    My relatively recent Mac (2018) doesn't have TOSLINK (neither in nor out).

     

    Yeah, that's why I keep this old Mac -- it's just too convenient. Like I mentioned, without such a Mac you'd need another interface that has a digital input. 

    Share this comment


    Link to comment
    Share on other sites

    This is getting really interesting. 

     

    I'm trying to figure out why the bit perfectness of files going USB out is spotty. For example, an HDCD album from Pearl Jam will illuminate the HDCD light for the first half second, then it will come on for another half second about a minute into the track. This lead me to believe there was some dynamic watermarking involved. Like it was an add-on file that was played simultaneously, rather than embedded into the music. If the watermark was a minute long and inserted dynamically, I could explain the blips of bit perfection. this is just total speculation on my part though. 

    Share this comment


    Link to comment
    Share on other sites

    On 6/18/2021 at 5:03 AM, new_media said:

    I think that an Apple TV could be used headless to stream Apple Music lossless to an audio-only rig, but you would need a DAC with an HDMI input.

    I also did some experiements with this. I plugged my Apple TV 4K (2018 model) into my NAD M10. I also have an iPad plugged into the same DAC via USB. I played a track that reports at 24/96 on my iPad. I then on my iPhone started the music app and chose to control the Apple TV remotely (not AirPlay). It reports on the iPhone as being 24/48 and not 24/96 as capable. I tried the same with a 24/192 track. If I quit the Music app on the iPhone the track continues to play on the Apple TV. If you AirPlay, the stream stops as it requires the remote source. A real cludge and limited to 24/48, but nevertheless better than AirPlay. All you would need is a Apple TV 4th gen and above 32Gb and a DAC that accepts HDMI.

    Share this comment


    Link to comment
    Share on other sites

    2 hours ago, Neil Lavitt said:

    If I quit the Music app on the iPhone the track continues to play on the Apple TV.

     

    I also determined that from Apple Music this remote control is possible also. I was also able to confirm that no data was actually coming into and leaving the Mac while music was playing. When Apple Music was quit on the Mac, the music continued on the Apple TV. This would appear to confirm the Apple TV is playing the stream directly and in lossless up to 24/48.

    Share this comment


    Link to comment
    Share on other sites

    3 hours ago, new_media said:

    The last several versions of MacOS have allowed an iPhone or iPad to be added as an audio source.

     

    Very interesting. Thanks.

    Share this comment


    Link to comment
    Share on other sites

    FYI - Apple just updated their Android app (again) earlier today

     

    https://play.google.com/store/apps/details?id=com.apple.android.music&hl=en&gl=US

     

    These guys should be coming from the beta version back in June

     

    http://www.headphoneclub.com/thread-745437-1-1.html

    UfrMTvRl.jpg

    d3c6OxUl.jpg

    dYVVMyIl.jpg

     

    Other than that, we'll definitely need one of those portable DAPs with Direct Transport Audio (DTA) in order to bypass SRC of Android. (e.g. Cayin / FiiO / HiBy etc.)

     

    As usual nobody was trying to confirm whether it's bit-perfect or otherwise.

    Share this comment


    Link to comment
    Share on other sites

    DUDE, I just checked this out again this morning and it's finally here

     

    https://play.google.com/store/apps/details?id=com.apple.android.music&hl=en&gl=US

    U2XOCqi.png

     

    What a bold claim, though it might seem to promise everything and deliver nothing.

     

    Unless we've got specific Android devices that could actually bypass SRC to begin with, what's the point of that particular statement mentioned above?

     

    Of course Apple might somehow manage to see the light and implement something that's similar to what we could get from USB Audio Player Pro etc.

     

    https://www.head-fi.org/threads/usb-audio-player-pro-uapp-24-and-32-bit-playback-ubiquitous-usb-audio-support-for-android.704065/page-54#post-12855034

    Quote

    Confused about the bit perfect option. I thought the point of the app was already to bypass any processing Android does and deliver the true output to the DAC. Now I don't know the difference between enabling the option or not, and nothing explains it. Frustrating.

     

    https://www.head-fi.org/threads/usb-audio-player-pro-uapp-24-and-32-bit-playback-ubiquitous-usb-audio-support-for-android.704065/page-54#post-12855207

    Quote

    It means that the audio that is present in the audio file is sent unaffected to the DAC. Of course this does not apply to compressed audio, like mp3, but holds for wav and flac and other lossless audio.
    Without the bit-perfect option enabled, all audio that is read from the file is converted to 32-bit floating point and then converted to the appropriate 16/24/32 bit integer format when sending to the DAC (after optional processing that you chose like EQ, balance, upsampling). This leads to an error for about 20 or so out of 16.7 million values that are 1 bit 'off'. Although you wouldn't hear that amplitude change, it might distort high frequency content a little. Most people won't hear it though.

     

    Share this comment


    Link to comment
    Share on other sites

    15 minutes ago, seeteeyou said:

    DUDE, I just checked this out again this morning and it's finally here

     

    https://play.google.com/store/apps/details?id=com.apple.android.music&hl=en&gl=US

    U2XOCqi.png

     

    What a bold claim, though it might seem to promise everything and deliver nothing.

     

    Unless we've got specific Android devices that could actually bypass SRC to begin with, what's the point of that particular statement mentioned above?

     

    Of course Apple might somehow manage to see the light and implement something that's similar to what we could get from USB Audio Player Pro etc.

     

    https://www.head-fi.org/threads/usb-audio-player-pro-uapp-24-and-32-bit-playback-ubiquitous-usb-audio-support-for-android.704065/page-54#post-12855034

     

    https://www.head-fi.org/threads/usb-audio-player-pro-uapp-24-and-32-bit-playback-ubiquitous-usb-audio-support-for-android.704065/page-54#post-12855207

     

    Interesting claim!

    Share this comment


    Link to comment
    Share on other sites




    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now




×
×
  • Create New...