Testing audiophile claims and myths

[Vamp898]

You are saying LDAC uses a lossless compression, but does Sony?

I just repeat what Sony says, i did neither invented LDAC nor developed an LDAC Codec

Here is the source of libldac: https://android.googlesource.com/platform/external/libldac/+/refs/heads/android11-d1-s7-release/src/

Maybe someone with more C knowledge than me can explain how and why it works.

But i did find conflicting information on the Sony website!

There are Multiple versions of the LDAC Explanation site (depending on which product you look at).

The version i know (which was also translated into English as i found) said "Identical to CD Quality"

The Version that is used for Active Speakers only says "Playback with High Quality"

So it seems that Sonys itself doesn't really know if its lossless or not.

Hmm... To be sure I re-encoded with flac 1.3.3 and best compression:

I took a deeper look on how FLAC works because my music does have much lower bit-rates (mostly. Some Bands as Eluveitie show similar results, but thats kinda it).

It seems that the structure of the Song plays a big role in how good FLAC can compress it.

It seems like the Sons i listen too are well structured (but the most complex songs i have/know are all in Hi-Res Audio, maybe i should convert them to 44.1KHz/16bit and check how well FLAC can compress them.

There are some tests on the Internet who recorded the output that came out of an DAC using LDAC and compared it to direct playback from the same DAC and mentioned that there is no difference.

One of those tests, and i think the most famous one, is this one: https://www.soundguys.com/ldac-ultimate-bluetooth-guide-20026/

 

[Davesrose]

Careful with the marketing blurb, as LDAC might be identical in "quality", but it would be impossible to be bit perfect and truly lossless as the max bit rate (990 kbps) is lower than a CD (1411 kbps).

I just repeat what Sony says, i did neither invented LDAC nor developed an LDAC Codec

Here is the source of libldac: https://android.googlesource.com/platform/external/libldac/+/refs/heads/android11-d1-s7-release/src/

Maybe someone with more C knowledge than me can explain how and why it works.

But i did find conflicting information on the Sony website!

There are Multiple versions of the LDAC Explanation site (depending on which product you look at).

The version i know (which was also translated into English as i found) said "Identical to CD Quality"

The Version that is used for Active Speakers only says "Playback with High Quality"

So it seems that Sonys itself doesn't really know if its lossless or not.


I took a deeper look on how FLAC works because my music does have much lower bit-rates (mostly. Some Bands as Eluveitie show similar results, but thats kinda it).

It seems that the structure of the Song plays a big role in how good FLAC can compress it.

It seems like the Sons i listen too are well structured (but the most complex songs i have/know are all in Hi-Res Audio, maybe i should convert them to 44.1KHz/16bit and check how well FLAC can compress them.

There are some tests on the Internet who recorded the output that came out of an DAC using LDAC and compared it to direct playback from the same DAC and mentioned that there is no difference.

One of those tests, and i think the most famous one, is this one: https://www.soundguys.com/ldac-ultimate-bluetooth-guide-20026/

Why not start with Wikipedia? Wikipedia says LDAC is both lossless and lossy (link):

LDAC is a lossless and lossy codec,[2][3] which employs a hybrid coding scheme based on the Modified discrete cosine transform[4] and Huffman coding[5] to provide more efficient data compression. By default, LDAC audio bitrate settings are set to Best Effort, which switches between 330/660/990 kbps depending on connection strength;[6] however, audio bitrate and resolution can be manually adjusted on Linux (when using Pipewire[7]), and on some Android platforms, Sony smartphones and Walkman devices at the following rates; 330/660/990 kbps at 96/48 kHz and 303/606/909 kbps at 88.2/44.1 kHz with depth of 32, 24 or 16 bits. When the codec is set to 16 bits/44.1kHz at 909 kbps (or 16 bits/48kHz at 990 kbps) LDAC can stream lossless audio that is identical in quality to (or slightly higher than) Audio CD or standard resolution uncompressed audio.

 

[sonitus mirus]

That still does not make a lot of sense. They are streaming a lossless source at a quality level that is identical or better. It is not bit perfect lossless and I am unable to find any direct reference from Sony stating such a claim. It is like MQA lossy. It might sound great, but it is technically lossy with regards to CD. But, for that matter, AAC 256 is identical in quality to any 24/192kHz file from an audible perspective.

 

[Davesrose]

That still does not make a lot of sense. They are streaming a lossless source at a quality level that is identical or better. It is not bit perfect lossless and I am unable to find any direct reference from Sony stating such a claim. It is like MQA lossy. It might sound great, but it is technically lossy with regards to CD. But, for that matter, AAC 256 is identical in quality to any 24/192kHz file from an audible perspective.

I would think LDAC can be like Dolby TrueHD: when you encode, you're encoding at the higher lossless bitrate, and there can be a "core" Dolby Digital track for devices/bandwidths that can't take it. If you're getting 990kbs it's at least very close to bit perfect with uncompressed 16 bits/48kHz (I'm less familiar with audio codecs, but modern video codecs improve quality with given bitrates quite a bit with each generation of specs).

 

[Vamp898]

That still does not make a lot of sense. They are streaming a lossless source at a quality level that is identical or better. It is not bit perfect lossless and I am unable to find any direct reference from Sony stating such a claim. It is like MQA lossy. It might sound great, but it is technically lossy with regards to CD. But, for that matter, AAC 256 is identical in quality to any 24/192kHz file from an audible perspective.

Its possible to distinguish an (CBR) AAC 320 from an 16/44.1kHz file.

Even with average J-Rock/J-Pop songs like オンリーワンダー from フレデリック its possible to spot differences, how should there be no difference from AAC 256 file?

Of course not with every file/song but with lots of songs even the measurements in Audacity say that there is an audible difference. That doesn't mean some is able to spot them, but they are there and they are audible depending on the listener and the setup.

I assume that not every Headphone/DAC/AMP combination is good enough, but i was able to spot those difference back with my NW-ZX507 and an MDR-Z7M2 and those are neither the top of the mountain nor ultra detail revealing.

 

[bigshot]

Its possible to distinguish an (CBR) AAC 320 from an 16/44.1kHz file.

No it isn’t. AAC is transparent even below 320. Do a level matched, direct A/B switched blind comparison and you’ll find out.

 

[castleofargh]

Its possible to distinguish an (CBR) AAC 320 from an 16/44.1kHz file.

Even with average J-Rock/J-Pop songs like オンリーワンダー from フレデリック its possible to spot differences, how should there be no difference from AAC 256 file?

Of course not with every file/song but with lots of songs even the measurements in Audacity say that there is an audible difference. That doesn't mean some is able to spot them, but they are there and they are audible depending on the listener and the setup.

I assume that not every Headphone/DAC/AMP combination is good enough, but i was able to spot those difference back with my NW-ZX507 and an MDR-Z7M2 and those are neither the top of the mountain nor ultra detail revealing.

Be careful here. Audacity doesn’t care about auditory masking, your brain doesn’t have that option.

 

[Vamp898]

Do a level matched, direct A/B switched blind comparison and you’ll find out.

I did exactly that, and i was able to spot the FLAC every single time with the above mentioned song.

There is a specific part in the song that starts at 02:36 where you can hear differences in the ringing of the snare first and than in the attack of the ride starting at 02:48

All i did was to wait for that part and i instantly knew if it was the AAC or the FLAC, simple as that.

This also matches exactly the values measured by Audacity where Audacity shows an audible difference at those parts (beside other parts).

I compared the (CBR) 320kbps AAC with the 48 KHz 24 bit Version of the song.

With the VBR 320kbps AAC (which goes up to 576kbps) those differences disappeared. I was no longer able to tell a difference with the VBR 320 AAC, but switching back to the CBR version, i was again able to tell the differences.

So there is an audible difference between CBR 320 AAC and VBR 320 AAC, so of course there is an audible difference in CBR 320 AAC to FLAC

 

[71 dB]

I did exactly that, and i was able to spot the FLAC every single time with the above mentioned song.

There is a specific part in the song that starts at 02:36 where you can hear differences in the ringing of the snare first and than in the attack of the ride starting at 02:48

All i did was to wait for that part and i instantly knew if it was the AAC or the FLAC, simple as that.

This also matches exactly the values measured by Audacity where Audacity shows an audible difference at those parts (beside other parts).

I compared the (CBR) 320kbps AAC with the 48 KHz 24 bit Version of the song.

With the VBR 320kbps AAC (which goes up to 576kbps) those differences disappeared. I was no longer able to tell a difference with the VBR 320 AAC, but switching back to the CBR version, i was again able to tell the differences.

So there is an audible difference between CBR 320 AAC and VBR 320 AAC, so of course there is an audible difference in CBR 320 AAC to FLAC

So does CBR AAC ruin the enjoyment of listening to the song for you, because two short parts of the song sound a little different?

 

[Blaithin]

I did exactly that, and i was able to spot the FLAC every single time with the above mentioned song.

There is a specific part in the song that starts at 02:36 where you can hear differences in the ringing of the snare first and than in the attack of the ride starting at 02:48

All i did was to wait for that part and i instantly knew if it was the AAC or the FLAC, simple as that.

This also matches exactly the values measured by Audacity where Audacity shows an audible difference at those parts (beside other parts).

I compared the (CBR) 320kbps AAC with the 48 KHz 24 bit Version of the song.

With the VBR 320kbps AAC (which goes up to 576kbps) those differences disappeared. I was no longer able to tell a difference with the VBR 320 AAC, but switching back to the CBR version, i was again able to tell the differences.

So there is an audible difference between CBR 320 AAC and VBR 320 AAC, so of course there is an audible difference in CBR 320 AAC to FLAC

It is possible to train yourself to identify the masking and artifacts consistently, but it's really not something you want to do unless you're being paid to do it.

I believe Amir of ASR has that kind of training, and even he's hesitant to recommend people undergo that process.

The differences between lossy and lossless are subtle unless you use "killer" tracks that will sound very wrong when turned lossy. The content of the recording matters much, much more. You can have all the bits at your disposal, but a crappy performance and/or recording is still crappy even in lossless format. But an excellent performance AND recording, you could compress it and STILL the "soul" shines through. IMO, If the essence of something is beautiful, no amount of corruption will tarnish that beauty.

I have a great example linked below:


Voices of Music Vivaldi Winter w/ Cynthia Freivogel.

The above recording exists only in lossy formats (Ive tried to buy it from them. No dice. Donated instead.), but it is bar none the most beautiful recording and performance of Vivaldi's Winter I've seen and heard to date. The church's acoustics coupled with that sublime performance never fails to bring me to tears. Im not even a classical/baroque music guy and the Vivaldi suite by these performers keeps me coming back.

I have no quandaries about not knowing whether I can hear lossless vs lossy. I prefer the ignorance of being able to enjoy the vast majority of the world's music. To give up variety and the ability to experience new talent and new music over a format..For the moot point of proving to myself that I've got golden ears? (I don't. I've gone down the same path as you)

It's just not worth it if I have to draw this much blood at the puritanical altar of "Sound."

There are just some lines not worth crossing imo

 

[sonitus mirus]

I did exactly that, and i was able to spot the FLAC every single time with the above mentioned song.

There is a specific part in the song that starts at 02:36 where you can hear differences in the ringing of the snare first and than in the attack of the ride starting at 02:48

All i did was to wait for that part and i instantly knew if it was the AAC or the FLAC, simple as that.

This also matches exactly the values measured by Audacity where Audacity shows an audible difference at those parts (beside other parts).

I compared the (CBR) 320kbps AAC with the 48 KHz 24 bit Version of the song.

With the VBR 320kbps AAC (which goes up to 576kbps) those differences disappeared. I was no longer able to tell a difference with the VBR 320 AAC, but switching back to the CBR version, i was again able to tell the differences.

So there is an audible difference between CBR 320 AAC and VBR 320 AAC, so of course there is an audible difference in CBR 320 AAC to FLAC

When I mentioned AAC 256, I was referencing Apple Digital Masters and what I recommend most people should be using when encoding to lossy for any reason.

https://www.apple.com/apple-music/apple-digital-masters/docs/apple-digital-masters.pdf

". A 256 kbps AAC file is encoded with a target bit rate of 256 kilobits/second. It utilizes Variable Bit Rate (VBR) encoding, which uses each bit strategically, dynamically allocating less data for simple sections and more data for complex passages."

 

[bigshot]

I did exactly that, and i was able to spot the FLAC every single time with the above mentioned song.

Try it again and lower the volume on the FLAC a hair before you encode it in AAC. It may be clipping. Some encoders boost the volume slightly when they encode, pushing hot mastered tracks a little into clipping.

The above recording exists only in lossy formats (Ive tried to buy it from them. No dice.

When you compare a lossless from one source with a lossy from a different vendor, you run into mastering differences. The way to tell is to take your own lossless copy and encode your own lossy copy.

Also, there is no reason to use CBR. VBR can't hurt. It can only help.

I wonder why it is that whenever I ask the info on killer tracks, it always seems to be a really obscure foreign CD that costs a lot of money to order from overseas. I just did that for another person's killer track. I'm not eager to do it again. Why don't domestic CDs at reasonable prices ever have killer tracks? I bet there it has something to do with expectation bias.

 

[GearMe]

...You can have all the bits at your disposal, but a crappy performance and/or recording is still crappy even in lossless format. But an excellent performance AND recording, you could compress it and STILL the "soul" shines through. IMO, If the essence of something is beautiful, no amount of corruption will tarnish that beauty.

...I have no quandaries about not knowing whether I can hear lossless vs lossy. I prefer the ignorance of being able to enjoy the vast majority of the world's music. To give up variety and the ability to experience new talent and new music over a format..For the moot point of proving to myself that I've got golden ears? (I don't. I've gone down the same path as you)

It's just not worth it if I have to draw this much blood at the puritanical altar of "Sound."

There are just some lines not worth crossing imo

Amen

Rather hear a song I love as a 128kbps MP3 on a boombox, than a pristine audiophile release of some yawnfest on a state-of-the-art system!

Nice performance...btw!

Bought a couple of their CDs

...I wonder why it is that whenever I ask the info on killer tracks, it always seems to be a really obscure foreign CD that costs a lot of money to order from overseas. I just did that for another person's killer track. I'm not eager to do it again. Why don't domestic CDs at reasonable prices ever have killer tracks? I bet there it has something to do with expectation bias.

So...Voices of Music is a non-profit based out of San Francisco. They sell their CDs for $9.99 - here

No Vivaldi - Four Seasons as of yet though.

Sounds like they're working on delivering a 4K video of The Four Seasons, are half way through, and would appreciate donations.

 

[bigshot]

I agree too. One of the best recordings I've ever heard was an LP of Fiedler's Gaeitie Parisienne that was recorded in 1952. I think it was the second Living Stereo recording made by RCA. The stuff that matters are the judgements made by the musicians and engineers, not incremental increases in data rates. I think sometimes audiophiles spend more time listening to the formats of their music than they do the music itself.

 

[Blaithin]

Amen

Rather hear a song I love as a 128kbps MP3 on a boombox, than a pristine audiophile release of some yawnfest on a state-of-the-art system!

Nice performance...btw!

Bought a couple of their CDs





So...Voices of Music is a non-profit based out of San Francisco. They sell their CDs for $9.99 - here

No Vivaldi - Four Seasons as of yet though.

Sounds like they're working on delivering a 4K video of The Four Seasons, are half way through, and would appreciate donations.

Yeah, I'm waiting for that CD. Thanks a ton for the link! The person handling their YT channel never linked that to me. Just said that their songs would be available on Soundcloud.

Also, yes, I'm considering donating again due to my overly enthusiastic streaming of their content.