POLL Lossy Streaming Sound Quality

This is going to be more for the sake of entertainment than anything else, but maybe it could prove actually useful to some folks out.

https://drive.google.com/drive/folders/1cRdub95IsecozY5Lp5naOeXMXbetMek8?usp=sharing

The link above contains two audio files that are excerpts from a really popular song that some might think is overplayed but that I still love. I'd be curious to know which between the two you prefer and what you think the difference between the files may be (if any exist).

Feel free to go into detail about what specific aspects have you preferring the one or the other or theories as to what the differences may be, but to keep things relatively on the level please do hide your responses behind a spoiler tag per below:

I got a few questions from folks asking how to hide their responses behind a spoiler warning the last time I did something like this, figured it may be useful to have this here: you just need to type in the text in yellow (mind the slash to close) and type what you want hidden in between.




I'll disclose what differences there are (if any) once we have a fair number of responses. Will also disclose methodology afterwards just in case anyone wants to nitpick because I DO wanna do this right.

Cheers, and have fun!

 

Spoiler

PC System in signature through Susvaras

A seems to have a better sense of space.
You can hear more of the room with the singer and she sounds more like she's singing in front of me. The drums/percussion have better separation. The fading crash cymbal near the end fades more naturally. And the guitar's distortion has more crunch.

But who knows, A could be the lower quality version with some Alterations and/or EQ and B could be an unaltered higher quality version. Both sound pretty good to me though and I'd be happy with either if I was stuck with one or the other being the only option I could listen to.

Looking forward to finding out the differences! Thanks for the fun experiment, and great song choice btw.

 

forgot to add my own impressions:

Spoiler

Modi Multibit 2 (oversampling) > Magni Piety > HD600

I prefer A myself, it's got a bit of a richer, more vivid sort of saturation effect going on plus just a more "alive" sense of dynamics to my ear. The slammier low end is a great deal up my alley. Didn't really notice much in the way of spatial cues other than B felt a bit flatter overall in terms of contrasts of depth with there being a whit more haze.

I feel B sounds more "reference-y", but it's less enjoyable to me by a fair amount if I'm really listening.

@EstrangedBaron It's a beautiful song that I hardly ever get tired of listening to. The music video is difficult to watch but really adds to the experience. I'll wait until there are a few more responses but I'll say not to worry and that I didn't do anything to either track other than trim the ends so this isn't outright piracy haha.

 

Spoiler

Yggdrasil MIB (oversampling) > AMB ß22 > Beyer T1.2

Will look like a clown if you'll reveal there's no difference, but I'm gonna say A. Biggest cue for me is the staging of the vocals. It seems off in B, as if there were some stereo / joint stereo fuckery going on. It's not coherent in B.

Listening some more, I think I slightly like the definition of the guitar in A more also. It sounds more raw, but only slightly so.

 

Spoiler

PlayPcmWin 4x oversampling -> Soekris dac1321 -> WHAMMY -> YH-2 / HD6XX

Prefer B. I hear more distortion in A, esp in bass & guitar, and more hash in the highs and more grain in her voice. Agree bass is slammier in A.

I agree with the previous comments about B sounding more 'reference', 'smoother', less 'energetic', etc.; but in my system and for my (old) ears this is preferable.

B also seemed to me to have more (better) instrument separation, even though the staging seems more compressed.

Edit: I should also disclose I'm probably an outlier wrt the music: never been a Cran-fan, I always found her singing annoying

 

Spoiler

Im just casually listening, and almost certain my chain is messing with this somehow - I streamed straight from the Google files on my phone via Airplay. One of those probably messes with the bit rate, sampling and/or format - but I still hear a difference. I suspect B is the higher bit rate of the two. It has more space around each musical element, and also has a cleaner sound overall. I prefer it today, but on another day I might prefer the more “cohesive” (I.e. blended together) version A, depending on mood.

rest of chain - Allo DigiOne+ -> Bifrost 2 OG -> Decware Zen Triode -> Zu DW MkII

 

This was fun, nice idea

Spoiler

It definitely sounds like A is more compressed. Soundstage collapsed, more harsh.

B sounds more fleshed out to me ....I wouldn't say smoother since there's lots 'O distortion in this song, but the voice is easier to pinpoint, the instruments are more distinct and everything just sounds more "right" to me.

Watch I'll be totally ass backwards...that'd make me laugh.

Edit: looks like I'm in the minority so far...now I am laughingl.

 

Thanks, @Lyander , what fun.

Spoiler

I prefer B, but to my ears and via my desktop chain (Macbook Pro -> iFi Zen DAC V2 -> Piety -> Grado RS2x) the differences are small. A seems to have a little more compression hash around the cymbals, B has slightly smoother vocals.

EDIT: To echo @Biodegraded , I find her voice annoying and the song incredibly grating. BUT! To me that makes it an excellent candidate for critical listening -- zero chance of emotional or musical engagement

 

Spoiler

love this song

preferred A... sounded fuller and more meat to the notes B had more air to my ears and seemed "sharp" - initial impression was that it had more detail, but after listening to each a few times, I'm not sure that it did... All that said, I was listening on my slightly better that crap AudioEngine A2+ computer speakers (with a cheap ass 8" sub that is under the desk)

 

Spoiler

B's imaging is more focused and is wider sounding than A. B is lossless to me while A is lossy

 

Spoiler

2020 MacBook Pro --> Koss Porta Pro and I vote A due to less mud

 

Spoiler

A lacked any depth, seemed short of micro dynamics. Didn't enjoy it.
B was more like what CD quality usually sounds like.

 

Spoiler

50/50, unbelievable! Just a few more responses to achieve sampling size beyond which the distribution is unlikely to change much (n>=20). Will require quite the streak to tip it to either A or B. Excited for the reveal, not asking to rush it.

 

Spoiler

Honestly I was waiting for about n=20 as well before disclosing the differences but what's genuinely amusing to me was how File A had a fairly pronounced lead to begin with after which File B overtook it only for A to match it again. This is fun stuff and does go to show that after a while it really does just come down to preferences

 

I completely forgot about this because of IRL stuff but for those curious (hopefully y'all vote before clicking on the spoiler below just to keep it fun):

Spoiler

File A (8 votes) is a rip of the song from YouTube Music whereas File B (7 votes) is a rip from Spotify Premium, both with things at their respective maximum quality settings. I can understand why Spotify might be perceived as lossless file relative to YTM since it does have more of a dynamically flat and less "saturated" sound to things. Both are more affordable in my country relative to others so it wasn't too rough to get the both of them, and I was curious to see which I'd prefer.

In my case YouTube Music was just more my vibe, but yeah there's just no accounting for preference. I honestly thought more folks would vote for Spotify since it's what I feel is a more "conventionally audiophile" sound, but hey good fun!

Methodology plus objective differences:

Spoiler

I basically set Audacity up to use my DAC as my recording device via loopback to cover the same segments of the song, after which I made sure to time-align them. Was interesting to see that the waveforms captured were slightly different between them, especially during the large dynamic swings where it seems like YTM has the more aggressive transient edges and more of a compressed effect while also having slightly greater dynamic range between peaks and valleys



Here's what happens when you invert the one and export, showing off the differences in terms of absolutes (wish there were a way to show which way the differences lean:



BLUE: YouTube Music
ORANGE: Spotify:

 

Spoiler

If the different bit rates didn't fool the conversion to WAV, there are a few dB differences in the sub-sonic frequencies - differences between the AAC and MP3 OGG algorithms, I guess? The rest look very close though.

File A (Youtube Music, originally 256kbps AAC):



File B (Spotify Premium, originally 320 kbps MP3 OGG Vorbis):



Thanks to @roderickvd for the corrections in red.

 

Spoiler

That's very interesting stuff, didn't think to look that far down! Might get one of my CD rips (I have recently started buying CDs because, somehow, this is cheaper than buying FLACs online) and convert to both 320 MP3 and 256 AAC to see if there's a similar difference! Wonder how much of an audible difference there'll be, if any, solely owing to that infrasonic difference

 

Spoiler

Thanks a lot for this. I would never have thought I'd given the nod to YouTube Music. Pretty sure that if you had told me, it being YouTube as well as lower bitrate and all that, I wouldn't have.

I'm assuming the mastering is the same, as it's pretty darn close.

Here's a spectrogram of file A:


And of file B:


When you compare the spectrograms full screen, you'll see that file B has more high-frequency content than file A. An earlier plot also showed A to have more low-frequency content than B. In some sense, that would give it a bit of a loudness tilt. correction: B has more LF too

Note that Spotify streams Ogg Vorbis, not MP3. There definitely is a difference between different codecs. I've never compared AAC and Vorbis before. But I've tried MP3 and Vorbis. There, cymbals sound different: Vorbis tends to sound more smeared out, with MP3 having more of a "twang". And though MP3 is ancient in comparison, I still tend to prefer it over Vorbis.

 

Vice-versa.