It's not every day you get to make a new master of an album you've already worked on – especially not from a completely new mix !

But that's exactly what happened for me recently.

Dan from Ember Rev had never been happy with the guitar parts on their 2019 album "From The Country To The City To The Sea" and recently decided to re-record them. At the same time he used the opportunity to make some changes to the arrangements and song structures that more closely match the way the band perform them live.

The album was originally mixed by Simone Silvestri, but for this new version Simone wasn't available and Dan asked Sam Inglis from Sound On Sound to handle the re-mix. Sam took quite a different approach, which he's written about in detail for a fantastic new article in the latest issue of the magazine. He also asked me to contribute a short segment talking about the differences I heard from a mastering perspective, and the different approaches I took when mastering each version.

You can read the full article here, (or sign up here to read the whole "Replica" issue for free !) and in the video above I demonstrate how all this sounds in practise, using the song "Ultramarine" as an example. You can download and compare the files for yourself, plus many other examples, here.

How much difference does mastering really make ? And perhaps more importantly, how much difference should it make ?

Take a listen and decide for yourself !

PS. You might also like to take a listen to Ember Rev's latest album, also mixed by Sam and mastered by me, here

I get quite a bit of push-back about the Loudness Penalty idea – and I think some of it is a simple misunderstanding.

In a nutshell, the "penalty" isn't about the numbers, it's about how the music sounds because of those numbers.

It's easier to hear (and see) than explain in words, which is why I made the video above, using example songs from this week's UK Top 20, including "Golden" from the K-Pop Demon Hunters soundtrack.

It's a great tune, but like most K-Pop the loudness has been pushed really hard, so I couldn't help asking - how would it sound with better dynamics ?

And in fact we don't have to wonder - I can demonstrate it to you.

Which is exactly what I do, along with two other examples.

Take a listen – can you hear the difference ?

I'm seeing this question all the time recently:

"I hit the LUFS target, but my mixes still sound smaller than the pros - WTF?!"

Which seems like a perfectly reasonable question, right ? LUFS measure loudness, so just match the LUFS to the appropriate target and you're good to go… why wouldn't it work ?

Well (as always) it's not quite that simple.

Yes, LUFS attempt to measure loudness, but it's just a number, like anything else. It has distinct limitations - and it makes for a pretty hopeless target.

The infographic above tries to show this, using a visual analogy. You might have seen the optical illusion it's based on before – squares A and B on the chess board are actually exactly the same measured colour, as you can see in the zoomed-in version on the left, where I've joined them together to make it even clearer.

But in the zoomed-out version on the right, it seems impossible to believe. Our brains look at the context and decide that the white square can't possibly be the same colour - the black square seems obviously darker.

But they really are – I've checked the values myself in Photoshop. Even the letters A and B inside the squares are the same colour...! And even though I know that, I still see them as different colours, just like you. Perception is complicated – remember The Dress ?!

And something very similar can happen with audio. Even though two sounds measure exactly the same LUFS, we perceive them as having different loudness.

There's a whole range of reasons this can happen, but the short version is:

Context

In the context of the shaded chess board, we perceive A as being darker than B.

In the context of an album or playlist, we may perceive two songs with the exact same LUFS as being different in loudness.

But here's the thing, though:

This doesn't mean LUFS is broken

I see loads of people experimenting with LUFS, and getting frustrated that it doesn't work the way they expect, and concluding that the measurement and even whole concept of loudness-matching is broken. Which unfortunately causes huge frustration, as Russ Hughes pointed out on the Production Expert site just this week.

And I completely get that. As I've said many times, aiming at LUFS targets when mixing or mastering is not a strategy that you're likely to find very useful.

But the measurement does work reasonably well, for a wide range of material – especially given its age and the fairly simple method used.

Try disabling Sound Check on Apple Music, or the default normalisation in Spotify to see what I mean – the loudness matching is far from perfect, but it's way better than nothing.

(And remember, over 80% of music fans listen with normalisation enabled and don't even realise, unlike most of us in music & audio production.)

So where does this leave us ?

I completely understand the frustration with LUFS, but expecting a simple measurement to account for the detailed psychoacoustics of musical perception is a pretty bigger ask. It might be possible using modern tech, but when LUFS was introduced those tools didn't even exist – and they needed to run on very low-spec devices.

In the same way, simply measuring the colour-value of the pixels in the chessboard image can't take account of the complicated way we perceive the image - but this doesn't mean that measuring the colour in this way is broken or flawed. We just have to be careful with how we interpret both results.

Maybe better solutions will come along eventually, but for the foreseeable future LUFS is what we have for audio – and it does a decent job, much of the time.

So my suggestion is to follow musical methods for determining the loudness of our masters rather than trying to use LUFS targets – and then test carefully to make sure it has the result you're hoping for, both with and without loudness-matching.

It's the strategy I've been using my entire career, and it still works – regardless of the LUFS you choose to master at. Higher and lower levels each come with their own challenges, but it doesn't really matter what LUFS level you choose to master at - what matters is how you achieve that level, and all the other decisions you make afterwards follow from that decision.

It's not how loud you make it, it's how you make it loud.

"Headlock" by Imogen Heap went viral recently and has gone platinum, 20 years after it was first released.

To celebrate the 20th anniversary, a re-mastered version of the album "Speak For Yourself" has been released. But does it REALLY sound better ? Or just louder ?

In this video I take a listen to find out, using "Headlock" as an in-depth example.

1. The level you choose to master at influences every other aspect of the way your music sounds
2. The final level of your master is virtually irrelevant, and music fans don’t care

These two facts are both true - even though they seem to flatly contradict each other !

I call this the ‘Loudness Paradox’, but what does it mean ? How can both of these statements be true, and what are we supposed to do with that information ?

The second idea is quicker and easier to explain, and I’ve talked about it many times. Whether your song hits -8 LUFS, -10 LUFS or only -14 LUFS has almost no impact on the way the majority of people hear it. More specifically:

• Online, over 80% of people will hear their music “normalised”, meaning that anything louder than -14 LUFS will be dropped down in level, so all three examples above will be heard at the same distribution loudness
• Even when normalisation is off, research shows it doesn’t actually affect people’s opinion of the way the music sounds - certainly not in the way we imagine. They genuinely don’t care (apart from being irritated by having to change the playback volume)
• On physical formats like CD and vinyl, people simply adjust the volume as soon as the music starts playing. They’re used to doing this, so the biggest impact of a louder (or more dynamic) master is just an initial “huh !” before setting the volume. After this, people don’t notice the level of the master

All that seems fairly comprehensive, so why did I say the level is 'virtually' irrelevant ? I’m just being accurate. There are still a few situations where the LUFS does have an impact:

• If your music is intended to sound loud and is mastered lower than -14 LUFS, several streaming platforms won't increase the level so it likely won’t sound as loud as the loudest material, even with online normalisation enabled.
  (Music fans still probably don’t care, though !)
• Many DJs prefer consistent loudness in the songs they play, to make mixing easier without losing energy and flow, and to keep levels high if they don’t have control of the PA volume.
  (Even in this genre though, I see plenty of variety in loudness - lots of stuff is mastered up at as high as -6 LUFS or more, but there are also plenty of mainstream club releases at -8 and even down to -10 or lower. DJs seem to cope with these differences though, so I have to wonder if it’s really as big a problem as people say…?)
• I’m told that record labels, A&R reps and PR people often assume that high LUFS are required, necessary and/or desirable – and will react dismissively to quieter masters. I have no idea how widespread this is, but it’s a common concern and understandable that people might want to avoid the possibility.

So OK, the final LUFS is often irrelevant, with those few specific exceptions that may or may not concern you.

But I also said that the level you master at influences everything else about the way the music sounds ! How can that be true, if the final level is irrelevant ?

The answer is that loudness is at the core of everything. 

• When you write a song, you decide if you want it to be quite and gentle, or exciting and loud
• Probably the most fundamental performance decision you can make is how loud the music should be at each moment
• The most important mixing decision is how loud each part of the arrangement should be in comparison to the others
• Adjusting EQ is simply a process of choosing how loud each sound should be at a particular frequency
• Dynamics processing is all about managing loudness

So everything about your music and it’s emotional impact is affected by loudness !

And then you get to the mastering stage, where the overall loudness changes our perception of the music - louder audio sounds brighter and bassier, so the loudness you choose to master at determines what the best EQ balance will be, how much compression is needed, how clean or distorted it sounds… everything.

So what’s the answer to the paradox ?

Both the statements are true, but they apply to different stages of the music production process. 

Loudness affects everything that happens in the music production process, up to the point of release.

But after release, the actual level stops mattering.

Does that mean that all the decisions, changes and judgements along the way stop mattering too, then ?

No.

Because they’re part of the audio, now, embedded in the sound. They're the things that made it sound the way it does.

If you decided to master loud, chances are it’ll have less dynamics, less low end and might be more saturated or distorted - which could be exciting or perhaps fatiguing. The same song mastered at a lower level might sound underwhelming if it were mastered too low and would have benefited from more density and intensity - or it might have more space, depth and variety that really brings out the very best in the material.

And that’s the real value of mastering - not just to make things loud, but to choose the perfect balance of dynamics and loudness, based on the level you’ve chosen to master at. 

It’s not how loud you make it, it’s how you make it loud

I'm excited to announce that we've released a major update to my Dynameter plugin – Version 2 allows you to get even deeper insight into the dynamics and density of your audio with:

• Powerful new analysis modes — Stereo, Mid-Side, and Multiband
• The new PSR Overview bar for instant visual feedback
• Fully resize-able vector graphics interface
• The unique new Peak Aware functionality — easily understand if limited dynamics are due to heavy processing or just a natural property of the audio

It's always nerve-wracking releasing something new, and this was no exception !

Will people like it, will they find the new features interesting and useful, will we have missed any bugs...!

But the feedback so far has been fantastic, which is great news !

A few people have asked for more details about the new features though, so we've just added an "explainer" video to the sales page which walks you through the highlights, plus a brief recap on how Dynameter works if it's new to you. You can watch it in full above.

And if you see this before August 15th, the introductory offer price will still be active - you can get 40% off using the code DYNA2INTRO at the checkout. For all the details click here:

Dynameter 2 by MeterPlugs - New Version

And if you're upgrading there's an even better deal – just click "click here to upgrade" underneath the red 'Add To Cart' button at the link above to get your exclusive discount code.

I hope you find Dynameter 2 useful - please let me know what you think !

Everyone is talking about AI, especially in mastering.

And there are certainly some exciting developments and interesting new products.

But can AI really take the place of a human engineer ?

I recently had the chance to test this idea for myself, using an actual mastering project, and in the video above I show you what I found, and the conclusions I drew from it.

Take a look, the results may surprise you !

I just watched an interesting video that Drew Oliver made for iZotope, where he notices a "peak level bump" in a recent release, and suggests it might be a deliberate production strategy.

And maybe he's right, but I assumed something very different when I saw what he was talking about – and in fact I can think of several other possible explanations, too.

I thought you might be interested, so I made a video explaining my thoughts. Take a look and let me know what you think ! And be sure to check out the original video - the section I'm commenting on starts @ 12:31

Almost 30 years ago (!) I mastered what I believe was the second ever album of 'generative music' to be released.

Around the same time, inspired by the experience, I bought the first – Brian Eno's 'Generative Music 1' – but was never able to play it properly.

...until now.

This is the (deeply nerdy) story of those two albums, of why I love them both, and how I finally managed to listen to Brian Eno's 'Generative Music 1' myself, exactly as it was originally intended.

And full disclosure – it's also at least partly an attempt to deal with my current case of EWS (Eno Withdrawal Syndrome).

Because last Friday was the ENO:24 livestream – 6 different versions of the ground-breaking generative documentary streamed over 24 hours, plus interviews, live footage and the art piece Nothing Can Ever Be The Same.

I have to admit I binged, catching 3+1/2 different versions of the film over that time, and loved every minute of it. As I said on BlueSky, the more I watched the more blown away I am by the skill of the editing team who put together all the different (possible) segments behind the scenes, and along with Brendan Dawes' generative coding, the results were amazing.

But now it's over, and I am bereft – so I've been delving into my archive of Eno posts to fill the void.

And at the top of the list is what might just be the rarest Eno album ever - Generative Music 1. Play the video above to hear an example song, and read on...

This album was Eno's first ever release of generative music, and since then he's pursued the idea, teaming up with Peter Chilvers to create a series of much more well-known generative music apps for iOS like Bloom, Scape and Reflection.

One of the remarkable things about Generative Music 1 though is the humble tech used to create and play it at the time. A 33Mhz 486 PC running Windows 3.1 with only 8MB would do fine, provided the PC in question had the KOAN generative music player – and crucially, a Soundblaster AWE32 soundcard.

And there's the catch. I bought the album myself soon after it was released, after working with Tim Didymus mastering the CD version of his album Float, which was also produced using KOAN software. Tim had been working with Brian, and I was excited to hear his new album.

Sadly, my computer didn't have an AWE32 card, and without it Generative Music 1 sounded... cheesy at best – commical at worst.

So the files languished unplayed on my hard drive. Reconnecting with Tim in 2016, he confirmed that the only way to play the album now was to build a dedicated 90s spec PC. I experimented with software emulators like VirtualBox but they lacked the critical AWE32 hardware support, specifically the EMU8000 chip that Eno programmed at a very deep level to achieve the results he wanted for this album. So the files continued to languish !

Until last week, when I came upon this video. AT the end of 2024, YouTube user Soft Automaton started uploading archival videos of 'Generative Music 1' being played back on exactly that kind of dedicated 'vintage' PC hardware ! Which is great, because it gives anyone the chance to listen to different plays of the album.

For me though, it just wasn't enough. Once again I started scouring the internet looking for a way to play my own files in realtime, and this time I found the 86box emulator. This insane open-source project emulates not only PC software, but the hardware as well. To the point where you have to choose the motherboard and graphics card separately, edit the BIOS settings, get the IRQs and DMA settings right... ah, the memories !

And crucially, one of the sound card choices is the AWE32 – and yes, it also emulates the EMU8000 !

Bingo, you might think - and yes it was, but only after more time & effort than I care to remember figuring out a configuration that was stable but slow enough that my venerable cheese-grater Mac Pro could handle the demands of emulating not only an entire PC hardware rig but also a realtime FM synthesizer...

And the great news is, it worked ! As you'll know already, if you've watched the video above. 'Generative Music 1' is playing in the background as I type this, in all it's 486 33Mhz emulated generative glory – and what's most interesting is that unlike Bloom or Reflection, separate plays of the same song can sometimes sound quite radically different.

So, is 'Generative Music 1' really the rarest Eno album of all time ? It's tough to say, but given that likely only 1000 physical copies were ever sold (I went for the cheaper download option), the files aren't online and even if you have them getting a hardware or software playback solution involves a considerable amount of effort, I'm guessing the answer is probably... Yes.

Now, time to dig out that 20-year-old laptop and see if I have the files for the three bonus tracks tucked away somewhere... #thequestcontinues

PS

If you've got this far you should probably check out Soft Automaton's Generative Music 1 playlist for yourself – and maybe also click this link for even more Eno nerdery from me. Hopefully they'll help cure your EWS !

Bonus – More Eno goodies !

Fantastic interview with Eno by Adam Buxton

And his classic animation of Eno in the studio with Bowie & co-producer Tony Visconti (doing more than people think)

Deleted track from 'My Life In The Bush Of Ghosts' – Qu'ran

Eno in conversation with Rick Rubin

In the studio for Reflection's generative creation process

Have you watched one of my videos recently and thought something like "this is all very interesting, but I'm not sure I can hear what you're talking about" ?

If so, then YouTube's new Stable Volume feature could be the reason.

In this video I show how it works, and how to disable it. Because when it's on, you'll never hear my videos the way I intended them 😕

I also explain why YouTube have added Stable Volume, and why it's not a complete disaster for everything we watch - but also why I'd like to be able to disable it permanently for my channel.