Rant on Audio Quality & Picking apart Formats

The origin of my rant sparked from this tread linked below. My reply was starting to build up a lot so… hence this new tread!

After putting myself to the audio quality test linked above, I felt there was not as much of a difference as I had pictured it in my head.

  • I felt that my success relied mostly on the genre of the song where there was more dynamic range. Listening to the way echo and reverb was presented helped me pick them apart but Coldplay and Katty Perry where just not giving me enough to be able to analyse properly (Also got impatient and maybe rushed a bit).

  • Nevertheless, I managed to score 4/6, wrong guessing the 128kbps Katty Perry and the 320kbps Coldplay.

Although I fully accept the fact that I cannot hear that analytically, I think the experiment is missing the point…

  • It does not matter as much if we listen to lossless in the instance where Windows 10 is going to fuck with the output (sample rate, bit-depth and all) when listening from the browser.

  • I believe the major shift in sound quality happens when using the Wasapi output on Foobar2000 or the “exclusive mode” on Tidal and similars on other platforms.

  • I’m willing to bet my ass out of the blue without having tested it that playing an 320kbps MP3 vs a FLAC on foobar2000 of a well recorded song would result in an almost undetectable difference (at least to my human ears, can’t promise for higher beings audiophiles [its a joke don’t be offended please]).

  • Also, I truly am of the opinion that a music file (MP3 or FLAC or any other) processed in the Youtube compression algorithm sounds so much worse than its original brother.

To resume, I’m firmly convinced that:

  • Listening to a quality 320kbps MP3 from a well recorded song & bypassing the Windows output mess with Wasapi OVER the Youtube playback encoded from a lossless file of that same song on a browser

    CONSTITUTES A BETTER IMPROVEMENT THAN:

  • Listening to a quality lossless file from a well recorded song OVER that same song at 320kbps MP3 (both the lossless and the MP3 been outputted the same way)

Putted even more plainly:

  • 320kbps MP3 + Foobar2000 + Wasapi output > Lossless + Youtube + Browser + Windows 10 output manager

    BIGGER IMPROVEMENT DIFFERENCE THAN

  • Lossless > 320kbps MP3

MIC DROP :facepunch: :raised_hand: :microphone: :arrow_down:

Please note that these are my opinions on the matter, NOT 100% TRUE FACTS.

Thanks to those who made this far: You’re AWESOME

3 Likes

Of course, youtube audio is about 128kbps aac quality so it’s going to be lower that both a 320kbps mp3 or a lossless file, and if that file was already lossy when it was encoded quality drops even further. It also depends on what resolution you use as well as with the lower resolutions audio quality drops further to save bandwidth

On the youtube version it’s going to be 128kbps aac, so it would make a bunch of sense that the mp3 is going to outperform on foobar vs youtube, using an exclusive mode or windows audio

I mean it’s pretty easy to tell the difference with decent gear and ample time, you can even boil it down to pattern recognition, if you learn what compression artifacts sound like and what lossy doesn’t have in terms of time domain information vs lossless, you can pretty much nail those tests with ease if you let yourself learn these things. If you are doing like background tests and not really focusing on anything them perhaps you prob can’t tell the difference, but with a good recording it’s rather apparent. Now this may be true if you have a really poorly recorded and mastered track, then it actually might be really hard to distinguish the difference since the track itself is just bad lol. I would rather take a well mastered track in mp3 than a lousy mastered track in lossless, but if you are using a good recording you should be able to tell the difference with some practice if you want to make it a skill lol.

I really feel like it would come down to the specific track at that point, if you have a mediocre mastered track I could see moving to mp3 from the browser as making a larger difference than the mp3 vs lossless, but I feel if you have a pretty well mastered and recorded track, there can be more difference moving to lossless than going from mp3 from youtube. It would also depend on the gear you have, if you are using entry level gear than the jump from lossless most likely isn’t going to matter as much (you should still be able to hear the difference but it might not be very significant), whereas in a higher preforming setup it’s going to end up being more significant

4 Likes

Oo yea!

By the 4/6 score i got.
Coldplay and Katty Perry was my “doom” as well. Just could not… failed.

2 Likes

Not if your browser supports Opus. The audio quality standard on YouTube currently is Opus @ 160 kbps, VBR encoding, and it can be damn near transparent. At least on the Opus webpage they claim it’s “pretty much transparent” at 128 kbps, and this seems to be supported by the listening test data on SoundExpert org: Audio quality of encoders at 128 kbit/s - SoundExpert
User rating 5 = “no difference” vs. the lossless sample, and you can see Opus gets a 4.56 (while every version of AAC already gets a score above 5, interestingly).

When you go to 160kbps Opus gets a 6.88, so it should be “beyond transparent”: Audio quality of encoders at 160 kbit/s - SoundExpert

Of course, since anyone can download audio samples and send in ratings, it’s just a general-public evaluation. Trained listeners might be able to hear differences up to ratings of 6 or 7, but the point is YouTube today is not what it used to be, recently uploaded videos can sound pretty damn good. The real risk of hearing bad quality on YT is in videos that have been on there for 10 years or whatever - those will still be at the bitrate YT was using back then, there’s no way to re-encode now because obviously YT hasn’t been hanging on to the originals all this time.

3 Likes

Good to know, so then the aac is only used if you stream a mp4 on there and opus for the webm version (if the video gets the webm treatment)?

Both opus and aac are going to sound pretty solid for a lossy file, I would say from the times I have tried opus vs aac, opus was preferable, but it’s still def not lossless quality if you are doing a serious listen

For sure it’s completely acceptable for casual listening, also on the content creation side of things creators have been putting more effort into actually properly recording and mixing audio for their videos which is nice as well (at least compared to years ago lol)

Looks like it. If I read the available streams for a recent video with youtube-dl, all of them are listed under extension=mp4 or webm, and all 3 Opus audio streams (at 50, 70 and 160 kbps) are webm. Easiest way to check per-video is to right-click the video area and select Stats for nerds (except that doesn’t tell you the bitrate).

1 Like

An iteration of this test had a Jay Z track on it that always tripped me up. I would consistently get 5/6 and miss the Jay Z one every time! But, that’s also a good segue into talking a bit about how audio compression works. The two primary ways of reducing an audio file’s [that space is very important! :stuck_out_tongue_winking_eye:] size is to remove all data above a cutoff frequency and also remove data that corresponds to quiet sounds that follow loud sounds, ie. sounds you barely register anyway and, theoretically, don’t miss when they’re gone. A truth about Hip-Hop and a lot of contemporary pop music is that it is rather simplistic; there isn’t as much melodic, harmonic, or instrumental complexity to it as there was in the 1960s-80s and part of 90s. There is less “quiet” after the “loud” to worry about. Now, I want to be clear…this is NOT a statement about artistic value or integrity, simply a statement that much popular music in the last 20 years has simply had less stuff going on in it. Less complex music is more compression friendly; it can have more data removed with less impact on overall quality and clarity. That is likely a big reason why many miss the pop and hip-hop tracks on that test.

2 Likes

I got 3 wrong, but on 2 of them I went for the 320kbps, so I’m not too sad about those. It’s the Coldplay one I’m still kicking myself about, because on first instinct I fkn picked the right one, just based on “liking it more”, but then I started overthinking it and looking for more treble reconstruction, and ended up picking the harshest one at 128 kbps. :face_with_symbols_over_mouth:

This is part of why I could never understand people who listen to metal and other harsh buzzy distorted music using “audiophile” gear. Like, how can you even tell you’re hearing it in “high quality” if it’s hella distorted by its very nature? :laughing:

3 Likes

It’s better to state the lossless format. Let’s add 24bit/96khz FLACs and DSD to the fight!

Storage is getting cheaper and FLAC files are easier to find. I have some lossy music but that’s because I couldn’t find the FLAC version, or simply the FLAC version doesn’t exist (like youtube music covers).

Oddly enough, I’m a big fan of DSD but don’t have a single DSD file. That’s because I EQ my songs. I even converted some DSD files to high-res FLAC.
Neutron Player converts music to DSD on-fly while playing. I tested various genres with and without EQ and DSD always sounded better than PCM.

I missed the classical and coldplay. Picked 320 on both. Not sure what that makes me and my speakers. 600M also i did some weed wacking over the weekend so not sure if that altered the test. :crazy_face:

Anyone that doesn’t get 6/6 is a big L and should get rid of all their high end gear since you’ve obviously wasted so much money on things you didn’t need. Offer me the first right of refusal on the items you sell, I pay $0.55 cents on the dollar. I bet you’ll be able to buy a great Beats HP with the money you get back.

:smiley:

3 Likes

This bring a compelling argument against WAV and DSD advocates where the handling of metadata is quite atrocious compared to FLAC and MP3.

Also, on a other avenue, I have seen many questionable arguments been made between WAV and FLAC advocates

edit because provided wrong link

No excuses there, I should have probably done more research on Youtube before involving it in my argument. Its not like I was planning to make that post. Words just wrote themselves and somehow I push the post button. How strange :thinking:

This is the reason why I didn’t get 6/6 listening from my pc → usb to DAArt Aquilla II → rca to THX AAA 789 → Sendy Aiva at volumes higher than recommended

Could not agree more with this statement, I got the “Shenmue Orchestra Version” soundtrack from 1999 and also some track from the in-game tapes that I swear sounds better than 50% of all my music that is lossless (FLAC, WAV and DSD)
Tho this might lead to the argument of nostalgia bias. anxious sigh*

Do people really claim Wav is better than Flac for music reproduction…?

There is an argument for Wav, it’s a simpler format, so the codec is trivial and likely available everywhere.
But both encode the exact same set of bytes losslessly, so neither has an advantage in playback quality.

1 Like

I know right, this is quite hard to think there would be such an argument in the Audiophile community. I really hope It was just some random thing I found and not a growing cult.

Nevertheless If I resume the thoughts I carried out from this article is that a conversion WAV → FLAC → WAV would not be a pure reversal.

To which I say that it’s a strange claim and don’t think can hold any weight in practice. In theory maybe a small bit of lost data caused from small approximations or errors could impact the double conversion. Though I believe its not likely to affect things significantly in practice to be even the slightest noticeable or even relevant.

A better subject of debate I carried out of it though is the idea that because FLAC requires more computing and calculations in order to be played opposed to WAV, this would have impacts in the audio quality of the playback, making WAV “the ultimate format” or something along those lines.

To that I shall not respond because I don’t think I’m even close to having the knowledge to be able to answer that. But… will still call it bullshit because…

This used to be the case long ago, but now in like all players at this point it’s pretty much identical and I really see no use to have your music in wav at all except if you plan to edit the audio since uncompressed audio is easier, otherwise for listening I personally don’t see a reason to use wav

2 Likes

How could there possibly be a difference if both are lossless? Different DAC architectures getting activated in each case?

I believe that’s the case.

Maybe it’s because of the lack of Dynamic Element Matching. Both DSD and R2R which many people prefer their sounds compared to regular dacs don’t have it while multibit delta-sigma modulators have it.

DSD section of the dac that I’ve tested is basically a lowpass filter. Neutron does the whole DSD processes.

You do know low pass filter means all you hear is bass right?

Lowpass filter is essential for DSD to remove high-frequency noise (+20khz).
As far as I know, all dacs with oversampling function have lowpass filter. These so-called digital filters that dacs have with names like sharp linear phase are all digital lowpass filters that happen after oversampling. BTR5 and Topping D90 have many of them.

What you said happens if someone use lowpass filter of EQ softwares and set the frequency to something like 100hz. In that case, you’d only hear bass. But I don’t use it.

1 Like