Video

Louder is more good

Interesting & informative podcast. There is a lot of good info on that Ozone article regarding platforms but at the bottom with the recorded sections of dynamic master spotify preview and the 'optimised version' I prefer the more dynamic version over the optimised. In fact I dont like what the optimised version does to the percussion by reducing it within the song (& it sounds a little more processed) and to me the more dynamic version just sounds better. And I can just turn the volume up if I want to hear it louder. So in my view the +3dBish optimised version is poorer not better! Of course this may also be subjective in that some may prefer it and others not, which ulimately shows you cant say or generalise that making a quiter section louder is objectively better because that subjective preference is still more an artistic decision for the musician or band. 00:12

Number 1

One caveat to mastering EDM specifically for club playback is that the PA system there is configured with massive headroom to allow for huge peak spikes from all the over-compressed low quality mp3s DJs bring on their USB sticks. It's not the same as playing those tracks on a tiny bluetooth speaker. Tbh at -8 LUFS integrated this master is simply too quiet. It'll stand out in a weird way in a 1h DJ mix next to 30 other tracks all between -3 and -4 LUFS. Most pros who prepared in advance would manage to compensate with the trim of the DJ mixer but a middle of the road DJ wouldn't do it perfectly and the mix will just sound uneven

Great Podcast!

i was watching this video for around 13min because the meterplugs facebook team said that the video will surprise me. first of all I'd like to say the video has an incredibly slow pace. it helps to think about the things you say, but i'm really impatient, which is why I have to stop halfway through hoping I got the main message and don't miss anything important. As far as I can see the argument of this video is that loudness matching songs gives you a useful perspective on the music, because when you hear how some dynamic songs come out with the same volume as denser tracks, that also makes them have a higher quality or something and we should all produce very dynamic music, because thanks to spotify and itunes we can do that without having to fear losing the loudness war. my problem with this attitude is that i just feel like it treats the problem in a superficial way. like we were only slapping limiters and clippers on stuff to boost loudness. no, we are also doing it for many other reasons. maybe a dense mix just sounds better in many cases. that scarred song for example sounded 1000x better than enter sandman, which just sounded weak. yeah, that was mostly due to the lack of lowend, just as you pointed out. but also the dense kick and snare, highly limited and clipped, really melted into the rest of the music in scarred, while in sandman the drums are just kinda there, but not locked into the other musical elements. they even had excessive reverb, which there would be no space for in a dense mix like scarred. so when you decide for density you don't only decide for loudness, but you decide for a dense sound, where things are really absorbing each other in an artificially tight space. that's why it sounds so sick. loudness matching will never make the weaker sounding dynamics magically superior. skrillex' music also lives from this super 2dimensional sound, that just refuses to show any sound on a different volume than whatever the biggest sound is in existence. that is all of brostep in general. only the most recent developments in dubstep try to make it more 3d with colour bass and botanica, where intense resonation in the sounddesign makes sure sounds have a bit of body and need space to breathe. but they don't do it to fit to the dynamic range trend, but because it sounds awesome and fluid. just like your song my helper btw. great mix as far as i can hear from the examples. but most music wouldn't need such an elaborate amount of dynamics where even a whole part has a different volume than a whole other part. in fact when i drive around in my car listening to a song like that i have to take one hand of the wheel to adjust the volume all the time, because it just makes no sense to have a part in there that just trys to be a little bit more demanding on my nerves, unless it's some sort of art piece or so. that's why I'd say it doesn't make a lot of sense to argue anything based on your track, until the track itself is mixed in a way where no one feels the need to adjust the volume knob anymore. i would have to listen to the full version to know more about how it feels, but when the loudness curve already says a whole part is louder, then idk.. probably i would turn it down. classical music also often has this problem. i love listening to all those beautiful piano arps, but when they bring in all those strings and it's suddenly 100x louder, idk i don't have any advantage from that. i just turn down the volume myself then and wonder why the mixing or mastering engineer didn't just do that for me

Hi Ian, when finalizing a track, from my understanding it's relatively common practice to trim the final output gain by -1db (as well as utilizing oversampling) to accommodate for potential intersample peaks/dsp overshoots as a result of DAC (lossy) and the potential for distortion/aliasing on certain systems. This may be a very simple to answer for those who have knowledge but my question is: When analyzing a track that has all ready been uploaded (let's say to Spotify for example) the track basically always hits up to 0db from what I can tell. If people are uploading with a trimming of about -1db or so, how come that doesn't show when looking at the track? Continuing with the example of Spotify, I know there is a Normalize volume setting but I have it turned off. Do streaming services like Spotify just bring the track up to 0db anyway, even without using the Normalize setting? If so that would explain why the tracks don't read at -1db at the loudest.

Given that AI is in the rapid process of ending music as a profession for about 80% of producers, worrying about loudness is not even in my peripheral thoughts anymore.

Thanks for another great podcast! Billie Eilish's latest is pretty dynamic by modern standards :D. I downloaded the HDtracks version, and that had anything from -9 LUFS Integrated down to -11,7.

See this guy is at -4.1? It sounds warm.

🤙🏻

Cant tell if magnepan or absorbers 11/10 great work sir

I prefer A by far. The vocals are more glued to the mix, the frequency balance is, for me, perfect, and the drums are punchy and clear. There's a super weird sound/room effect with B's drums, like it's really bad quality, like I'm listening to an mp3 at 64 kbps haha... And B's vocals seems to slightly lack body, and it's is the same feeling of "low quality" as the drums. Just my personal opinion.

Just listening to it on Spotify. Very interesting. Especially because I sometimes have to help musicians to have their material transferred to a pressing plant at a low budget.

Atmos binaural is processed poorly by CZcams. Listening to Atmos content here is like listening to FLAC file on a potato. You won't be able to appreciate it

Fantastic Vid.. Subbed...

I'm feeling like it's two different masters. Streaming version less limited/compressed, and CD version compressed that 3.3db further. I guess what might be confusing is that they may not be using the CD audio for the music video anymore. Knowing that most people will watch the video through CZcams and others, they went with the higher dynamic streaming version? Not sure if they were EQ'ed differently or if that would have been a side effect result of the limiting.

Wouldn't it be a more objective comparison if you had versions of the same song mastered with different LUFS and then matched? I have another question: With loudness normalization applied by the streaming platform, would the peak levels interfere with the penalty applied? For example, would a master with TP of 0.3dBFS be penalized more than a master with TP -1.0dBFS if they have exactly the same integrated LUFS?

LU/LUFS measurements are a massive improvement over any previous attempts to characterise perceived loudness. However, I don't see that the calculations fully accurately reflect perceived psychoacoustic loudness based on persistence and frequency bands and the repeated activity within that band. With Skrillex the continued insistence at the most sensitive frequency bands does, it seems to me, make it feel louder than LUFS indicates. No doubt LUFS has a way to go yet. Speaking as a sound guy and as a psychologist...

What a great demo of the Loudness Penalty Studio software..... I need to get that!

As far as I observed this for myself, the sum of the information/energy across the entire spectrum over time is also decisive for the perceived loudness. I recently made two songs and once again noticed that one song sounds louder than the other, even though they have the same LUFS value. The only difference I could tell is that one song has more total information spread across the frequency spectrum. I don't know if it would help to factor in the total information/energy per channel over time to get a better more reliable result! But maybe I'm completely wrong. 😅🤷‍♂

I’m not sure this helps, but I would say the greatest way to test this is on good, well mastered tracks that sound amazing, regardless of their final LUFS level. Take the first Keane album, made during the madness of heavy limiting. It still (to my ears) sounds like a beautifully mastered album. Perhaps there was some tricky multiband compression going on to achieve their sound, but it sounds ok. If you then normalise down their album to a reasonable level, you can still appreciate the way it was mastered. If you listen to ‘Broken…’ Metallica album it sounds overly squashed no matter what happens at the distribution end. Enter Sandman sounds better but not a lot. So, to summarise, I think the effect is entirely dependent (as you said) on EQ and the quality of the actual master. IMHO. :)

I think Bangarang sounds louder than your master on my phone. Maybe it's a playback system thing too?

The difference is so insignificant that you might as well master to the louder lufs for the 17-20% of people that do listen with loudness normalisation turned off.

Very interesting. I was convinced of that. However, a few colleagues are concerned that people who listen to the songs without normalization will feel that some songs are not loud enough. And as I understand the free Spotify tier does not apply optimization.

Enlightning

Man Why don’t Youjust let the Sia Song without the Match Mode 😡 didn’t fit your narrative?

A change of 8 dB to 12 dB is not a '50%' increase. dB is a logarithmic scale. The percentage difference insofar as it is meaningful is about 15% .

The author admits he's not aware of how the stages of cutting the disc from his inappropriate master may affect the dynamic range of the vinyl master but appears to imply that any changes must be down the DR meter's inability to do this correctly. I suggest several alterations would need to be made to the brickwalled file before it can be cut. Firstly it MUST be equalized according to the RIAA curve which considerably reduces the bass and considerably increases the treble. To considerably increase the treble on a file that's already brickwalled will require the overall volume to be taken down a notch or two first. If the CD master has very heavy bass (very likely) then this may require attenuation beyond that demanded by the standard RIAA. curve. Very loud bass cannot be tracked properly on vinyl when it's having to track high frequency sound at the same time. There are non- negotiable limits as to how vinyl can be pressed and remain properly trackable by a stylus that absolutely don't apply to CD. This and any other EQ changes could certainly (legitimately) affect the final DR score.

Unbelievably bad practice to supply a brickwalled master for a vinyl release.

The DR meter works perfectly fine for vinyl rips that have been equalized according to the RIAA standard curve for disc masters and where the digital files comprise at least one side of an lp that has been normalized to 0db or close to full scale BEFORE splitting into tracks. The problem is that it's not always possible to determine the process used by those submitting their results.

Mp3 of a reasonable bit rate have never produced significant DR differences in comparison to the lossless file in my experience of ripping thousands of CD s.

Great informative video..

Ian. I’m so glad you are still making videos and educating us on matters such as this. Appreciate this. I will no longer judge a waveform entirely if it looks like a pencil crayon. 😂

Great video, thank you.

Version A sounds waaaaaaaaaaay more "open" and alive

Many thanks to you both - fantastic episode !! I'd love to hear a follow up episode :)

Very informative. Thanks Ian. My favourite reference tracks on cd vary from -9 short term to -14. Your -10 short term for recommendation is helpful.

I’m much simpler way than this video is just turn your whole mix down by one or two DB. The volume wars are over people. Let there be dynamics and variety of volume in your mixes. It’ll sound more human and natural.

Sooo, when is Ep 2 dropping?? Hehe

I very seldom comment on CZcams videos but was compelled to on this episode. Absolutely awesome and fascinating. 11/10 and a big thank you.

Great podcast, I love the part of the recorded vinyl hiss. I believe that electronic noise gives a natural sense of sound that is not present in a digital system. Just like the universe has background noise, so does nature has its natural noise. Like the saying “it’s so quiet that you can hear the grass grow” For me analog components have that naturally that’s why I love working with analog equipment. Knowing how much knowledge and engineering experience goes into producing a record makes me appreciate listening and playing vinyl even more. Really wonderful podcast, I enjoyed it a lot and definitely will listen to it again someday 👌🏼

I definitely want to hear an other episode of this, I can listen to this for hours. Really great podcast with great timing because tomorrow my new record player will arrive and Sunday I’m going to a big vinyl market 😊

15 minutes in and already learning so much, that is so cool, thank you!

So the Sia track is 5 lufs louder than uptown funk, and you turn it down by 5 dbfs? What are you doing, sir?

I totally agree with this video. And here's a bonus topic that I've seen being debated here and there: Zooming in on the waveform and noticing that the waveform isn't symmetric - even after making sure there's no DC offset. Typically they see this with human voices or things that have been through vacuum tube gear. Now, as much as I understand the technical implications and potential effects on max output levels, I think the importance waveform symmetry is being exaggerated quite a lot. Or am I wrong?

Best bit of mixing information i ever heard was in a Dan Worrall video where he said that he preferred channel strip style EQs because you have to completely rely on your ears to hear the changes you're making. Crazy how easily we our ears are fooled by our eyes.

You should not have noise floor issue with 8 bit resolution if the 24 bit signal is converted to 8 bit logarytmick scale. I such case you will have full dynamic range with out audiable noise floor.

You also have to consider the zoom you are using. I always look at waveforms with a zoom of 4-5 mins of length so I standarize the process and then I can have a better guess of their dynamics. However I know the point of the video is to show that its more important to use our ears instead of the picture :)

In a previous video I was sandwiched between Skrillex and Daft Punk and now between Led Zeppelin and Porcupine Tree - I'm living the dream! I only tend to look at waveforms to get ideas about arrangement and locate the main sections easily in reference tracks.