Loudness war: what about it now? You Tube experiment.

[ct1]

 

Hello
I was with a friend of mine last week and he made a curious experiment before me. He mastered a track and compressed/limited it enough to be able to go as low as -6/-7db Lufs. Nothing unusual. He uploaded it on YT and got his video ready in no time. Right-clicking the video showed the normal Lufs value as “content loudness”=8db, meaning his track was indeed -14+8=-6. He then bounced the same song (compressed in the same way) but at -14 db Lufs and uploaded it too. This time around, YT said “content loudness=0”, which again is normal. I was amazed to hear NO DIFFERENCE WHATSOEVER in volume between the two tracks played by YT. Whether played on loud parts or quiet ones: absolute same volume level from the speakers. The -14 track was of slightly better sonic quality probably because the audio file had been left “more untouched” than the -6 track.
Then, my friend downloaded the compressed files from YT and each was their own: YT’s -6 was indeed a -6db file and YT’s -14 was a -14lufs file.

Many of you may explain this but I can’t. If YT doesn’t touch the Lufs level of the file, then does  YT’s Player do the trick and apply -x db of constant offset (with no compression to speak of, since louds and quiets sound the same between -6 and -14 ??) to any track YT analyzes as being of content loudness=x ??

Of course, the consequential question is: does uploading files at |LUFS| ≤14 make any sense now ??

Thank you for reading. Waiting for your answers since I believe I may be missing something.

 

[des99]

Most of these services attempt to balance out the playback volume. If you submit files that are louder than their reference documentation suggests, then they'll just turn it down to match their desired average and to give the listener a more consistent experience. If you submit files according to the required spec, then they won't have to adjust anything in playback. In short, if you won't get it to the levels they prefer, they'll do it for you.

It's always good to check the requirements and get in the habit of submitting files to the required spec, in general, for any delivery of files, rather than just ignoring the spec and hoping they'll sort it out at the other end, imo.

[ct1]

I know this.

The problem is that we've all seen on YT and elsewhere that audio files are generally uploaded way louder that -13 <-> -15. Go to YT, and see the stats of any ""professional" or "big name" video (including the most recent ones). You will see that the content loudness value is +4, +5 sometimes +6db, which shows that professional studios DON'T WANT to upload within YT's guidelines. Until now, it seemed that doing what they do was a "winning loudness war". Not as much now given the above mentioned experiment ??

Hence my question...

[Holger Lagerfeldt]

On 12/5/2023 at 5:56 PM, ct1 said:

 

If YT doesn’t touch the Lufs level of the file, then does  YT’s Player do the trick and apply -x db of constant offset (with no compression to speak of, since louds and quiets sound the same between -6 and -14 ??) to any track YT analyzes as being of content loudness=x ??

Of course, the consequential question is: does uploading files at |LUFS| ≤14 make any sense now ??

Thank you for reading. Waiting for your answers since I believe I may be missing something.

YouTube, like many other streaming services, does lower the loudness of a loud file by applying a static gain offset. This means e.g. -6 LU in your example.

This is done non-destructively by first measuring (after upload), then writing the offset into the meta data, then reading the meta data and applying the static attenuation during playback.

Uploading a file louder than -14 LUFS (I) makes sense if you like the sound of a less dynamic master - what it brings to the table in the form of sonic changes, density, and additional saturation.

Also, two masters, one made to target exactly -14 LUFS (/I) and the other say -9 LUFS (I) won't necessarily sound equally loud when played back at -14 LUFS (I). This has to do with a variety of psychoacoustic phenomena, including saturation.

The loud master might actually still sound louder than the quiet one, even though they're both played back at -14 LUFS (I). Or the louder one might sound equally loud, but too squashed. It all depends on the song and how it was arranged, recorded, produced, mixed, and mastered.

Furthermore, the way an integrated LUFS reading is measured includes a relative gate. This gate can negatively affect the loudness, fractions of a dB on the whole song, when it's not triggered. For example, a very dynamic arrangement with very loud choruses and extremely low breakdowns is usally penalized by an integrated LUFS measurement because the low sections are simply ignored by the relative gate.

Ironically, very dynamic pieces suffer the most from this kind of loudness measurement. LUFS (I) is better for extended program material like TV shows, not short pieces of music.

Whether someone prefers a more dynamic or a more compact sound is a matter of taste. Certainly a lot of electronic music and rock sound subjectively better to many when it's slammed a bit - a lot more than -14 LUFS (I).

If you pesonally can't hear the difference before and after normalization (relative to the original master sound) then that's not because the sound isn't different.

YouTube's guidelines are well-meaning, but just like Spotify's guidelines they're missing the point: it's a matter of taste, not a matter of pros ignoring (artistically arbitrary) guidelines out of spite.

Finally, the label usually doesn't order a variety of masters, but a single version for streaming/download (in 24 bits + 16 bits, perhaps one with a higher sample rate if the source file was HQ), but loudness will be the same.

Occasionally we'll deliver an Apple Digital Master, which has or used to have, specific demands for lossy encoding clipping, but other than that the same version goes out everywhere.

Streaming services range from no loudness normalization all the way down to around -16 during default playback, depending on the service.

It also depends on whether you use the free or premium version, your preferences for the particular service (-19, -14 or -11 in Spotify Premium - or no normalization), the type of device you're streaming to (via Airplay e.g. at around -18 with Tidal) - or no loudness normalization even when it's on, if you're using Spotify Connect. That's because the offset meta data isn't passed on via Spotify Connect, instead the device ignores this as it doesn't have the built-in limiter Spotify has. The limiter is always-on, but only engages with the "loud" setting of -11 LUFS (I) in case the static offset would otherwise cause (near) clipping. There's no limiter in the web version, PlayStation version, and with Spotify Connect.

For all of these reason it makes no sense sticking to arbitrary guidelines. We'll continue to master at the level we think it sounds right. The loudness war is still ongoing, but I think it's dawning on at least some people, that pushing the loudness for loudness' sake is usually a bad idea.

Broadcast is another thing, it makes a lot of sense to ease back on the clipping and limiting there, which is why I often make a dedicated radio version.
 

[ct1]

Thank you for this extended answer the details of which I cannot always understand because of my limited knowledge. The varied specs arbitrarily applied by streaming audio/video platforms make the process a bit of a headache though.

[Holger Lagerfeldt]

Indeed, which is why you can simply reduce everything down to this:

On 12/7/2023 at 4:22 PM, Holger Lagerfeldt said:

For all of these reason it makes no sense sticking to arbitrary guidelines. We'll continue to master at the level we think it sounds right.