Audio Quality & Loudness Advice

[Rough Slope]

Hi all,

Grateful for some thoughts and guidance on preparing a track for a local 4-piece rock band to use as they try to raise their profile, get some local radio play and a presence on Spotify etc.

I've been asked as a favour to take a listen to a track recently recorded - the band are not overly happy with the mix and, although I don't know the full details, they've been advised it is not well-suited for digital radio play. They shared some .wav files and I've been having a play with the mix to address some issues as I hear them there, but I was interested once I've done with that about how best to provide them with a file in a format and of a quality they can get some use out of.

The track is in proudly 80s hard/glam rock style (absolutely no quiet parts in the track whatsoever!), and the files provided were 24bit/44.1kHz, and I converted them/imported them into Pro Tools at 24bit/48kHz. I am no expert at mixing and far less mastering, but I've got a mix I'm generally happy with, and there is no clipping on the master meter. I had a look at the loudness of the mix, and it comes out at -19 integrated LUFS/momentary -14, and true peak -4.9dB. I ran it through Ozone 11 mastering software using a 'Rock' preset with settings untouched, and the result is a track which sounds alright on my monitors and in the car and clocks in at -10 integrated LUFS/momentary -5 and true peak 0dB. I've seen varying guidance on line for this from different sources but it would seem (rather simplistically) very much on the loud side. I was planning to export at 24bit/48kHz - .wav and MP3 format.

Would welcome thoughts on audio quality, loudness and format to provide the group something usable.

Thanks very much,
Neil

[Drew Stephenson]

So the short answer is that the track should be as loud as it needs to be.
Which doesn't help in your case.
-10LUFS might be fine if it's momentary, but less so if it's integrated. Your post isn't clear which one it is.
If it's the former, that's a perfectly reasonable level for a rock track to be hitting on streaming services. If it's the latter, then it's going to be turned down 3-6dB depending on the platform. That might not be a problem either. There's an awful lot of professional stuff being put out that's waaaaay hotter than that.
If there's no quiet parts in the track I'd at least try to make sure there's some dynamics in the mix though.

[The Elf]

The numbers you've given seem extreme at both ends. And I wouldn't be going anywhere near 0dBFS if it were my master.

I really wouldn't go throwing a 'hit and hope' at it with Ozone presets.

The important thing is to create a balanced mix that works for the song, then master it for the target format. 'Loudness' begins with the mix. If the band are serious about this they need to get an experienced engineer on the job.

[Rough Slope]

Thanks for the reply Drew - appreciated.

-10LUFS might be fine if it's momentary, but less so if it's integrated. Your post isn't clear which one it is.

I think it is. It's -10LUFS integrated, -5LUFS momentary.

All the best,
Neil

[Rough Slope]

Thanks for the reply too Elf.

I really wouldn't go throwing a 'hit and hope' at it with Ozone presets.

I've no intention to. I'm trying to learn something and was interested in seeing what the loudness result would be as a point of reference.

If the band are serious about this they need to get an experienced engineer on the job.

I'm sure they'd love to, but they are young lads with their own songs and no more money right now, and I'd like to help them for free - if I can.

All the best,
Neil

[James Perrett]

-10LUFS isn't actually that excessive these days. I've had clients expect me to go to -8 although I managed to convince them that -9.5 was more appropriate for the material. If I'm not specifically asked for a level, I tend to find around -11 or -12 suits most of the rock material that I work on.

Drew's songs probably wouldn't work at that level but that doesn't mean that -10 isn't appropriate in this case. As always, let your ears be the judge. Provided you have a good monitoring system then you should be able to hear when something is wrong - but if you don't have decent monitors then you should really persuade them to go to someone who has. There are a few mastering engineers on here who don't cost a ludicrous amount.

[The Elf]

the band are not overly happy with the mix and... they've been advised it is not well-suited for digital radio play.

Both of these things need addressing. The latter may be a five-minute job for mastering, but the former needs a mix engineer.

So you need to begin with the mix. Perhaps upload a section for us to hear?

[Drew Stephenson]

I think it is. It's -10LUFS integrated, -5LUFS momentary.

All the best,
Neil

Sorry, reading fail in my part. I also missed the 0db True Peak as well. Doh!

On the former, as James says, that might be right for the material but it will be turned down by most streaming services. Have a look at the Loudness Penalty website to see how that'll affect it. https://www.loudnesspenalty.com/

On the True Peak side, I'd give yourself at least 1dB of headroom there. The limiter in Ozone is easy to set for that and most likely won't have any perceivable affect on the loudness or the material.

[RichardT]

In terms of format, for streaming services, either 44.1 or 48kHz is fine and 24 bit is fine too.

A peak of -1 dBFS is generally OK, though for very loud tracks, some lossy streamers such as Spotify recommend leaving more headroom.

For loudness, it might help to ask the band if there are tracks they would like to use as a reference. Some rock tracks are limited to within an inch of their lives these days, and some are not.

Bear in mind that mixing can go well beyond balancing levels. So, with the band's permission, you could think about more radical things, if the track needs it.

[The Elf]

Time I did a good deed this week (and I'm passing on one done for me)...

If you can get the material (as a Cubase Project folder, or as time-aligned full-length stems) to me I'll do a one hour mix, If they don't like it then nothing spoiled. If they do like it then drop a few quid to my charity of choice.

If that works for you then drop me a PM.

[Hugh Robjohns]

...and clocks in at -10 integrated LUFS/momentary -5 and true peak 0dB.

The LUFS figures sound a bit high to me... but without hearing the actual material that's just an impression rather than fact!

However, the True Peak level at 0dBTP is a concern from a technical point of view.

The idea of True Peak metering is to try to catch intersample peaks. These can occur when the analogue audio waveform is reconstructed from the digital samples, effectively creating an output level higher than the individual samples... and so if the digital samples are close to, or at, 0dBFS the resulting output waveform can exceed the maximum output level of the D-A converter, otherwise known as clipping! Not nice...

The meters on most DAWs are simple sample-peak meters, which just show the highest amplitude value within a group of samples, and thst is usually lower than the true amplitude of the reconstructed waveform.... which is a problem for mastering engineers.

True Peak metering is a good attempt to overcome that problem — but it's not perfect —and because of that the very highest allowable value for True Peak is -1dBTP — NOT 0dBTP.

Let me explain...

True Peak metering works by oversampling the audio data, calculating intermediate samples between the originals. The higher the oversampling ratio, the greater the number of intermediate samples and the more accurately they describe the reconstructed waveform's amplitude.

Logically, the most accurate True Peak meter would have an infinitely high oversampling ratio because the greater the ratio the smaller the maximum possible error between sample amplitude and reconstructed waveform amplitude.

For example, the maximum amplitude error between sample amplitude and reconstructed waveform at x16 oversampling is just 0.05dB. At x8 oversampling the potential error increases to 0.2dB. And at x4 oversampling it can be as high as 0.68dB

But, a very high oversampling ratio is computationally very expensive, and DAW designers would rather use their CPU cycles for something more impressive than a peak meter!

Moreover, as the numbers above reveal, there's always a potential error margin even at quite high oversampling ratios, and we will always need to accommodate that potential True Peak under-read by imposing a small headroom margin.

So a compromise is required, using the lowest practical oversampling ratio combined with building in a suitable headroom margin to accommodate the worst case amplitude error under-read.

It makes sense to use a 1dB margin as the smallest practical division on an audio meter, and that would make using a x4 oversampling ratio viable, minimising the CPU hit.

And that's what has been defined in the BS.1770 /EBU R128 standards: x4 oversampling with a maximum permitted True Peak level of -1dBTP.

In this way, if the True Peak meter suggests the value is -1dBTP, but has a maximum possible under-read of 0.68dB, we know the true reconstructed output can't be any higher than -0.32dBFS, and therefore is absolutely guaranteed to still be safely below clipping.

If you allowed metered true peaks to reach 0dBTP the reconstructed waveform could hit +0.68dBFS... which means potentially clipped inter-sample peaks — the very thing we've been trying to avoid.

Sorry for the long-winded explanation... but it's important to understand why the True Peak meter isn't really true peak, and why the -1dBTP margin is absolutely critical to making the system work as intended.

Streaming services all reduce the level of content so that True Peaks don't exceed -1dBTP before considering the LUFS level. So, by mastering content to hit 0dBTP you've already imposed 1dB less volume through the streaming platform!

[James Perrett]

On the former, as James says, that might be right for the material but it will be turned down by most streaming services.

It will only be turned down when it is part of a playlist. Individual songs will often play at the higher level. However, the exact policies seem to be changing regularly so things could change next week.

The number one priority should be to get a mix that you are totally happy with. Don't even think about mastering, let alone releasing something, until you are happy with the mix.

[Rough Slope]

On the former, as James says, that might be right for the material but it will be turned down by most streaming services. Have a look at the Loudness Penalty website to see how that'll affect it. https://www.loudnesspenalty.com/

On the True Peak side, I'd give yourself at least 1dB of headroom there. The limiter in Ozone is easy to set for that and most likely won't have any perceivable affect on the loudness or the material.

No worries, and thanks for the link/guidance - very helpful!

[Rough Slope]

In terms of format, for streaming services, either 44.1 or 48kHz is fine and 24 bit is fine too.

A peak of -1 dBFS is generally OK, though for very loud tracks, some lossy streamers such as Spotify recommend leaving more headroom.

For loudness, it might help to ask the band if there are tracks they would like to use as a reference. Some rock tracks are limited to within an inch of their lives these days, and some are not.

Bear in mind that mixing can go well beyond balancing levels. So, with the band's permission, you could think about more radical things, if the track needs it.

Thanks Richard, very helpful. I did get a few reference tracks from them - a heady mix from the early 80s to present-day, and quite a range between them. All noted of course about mixing. All the best.

[Rough Slope]

Time I did a good deed this week (and I'm passing on one done for me)...

If you can get the material (as a Cubase Project folder, or as time-aligned full-length stems) to me I'll do a one hour mix, If they don't like it then nothing spoiled. If they do like it then drop a few quid to my charity of choice.

If that works for you then drop me a PM.

That's an incredibly kind offer, thank you. I'll check with the lads but I can't imagine they would be anything other than keen to take you up on that.

[Rough Slope]

Sorry for the long-winded explanation... but it's important to understand why the True Peak meter isn't really true peak, and why the -1dBTP margin is absolutely critical to making the system work as intended.

Thanks very much Hugh, and of course no apology necessary. This is incredibly helpful and I feel much better informed by this and all the other contributions. Much appreciated.

[Rough Slope]

On the former, as James says, that might be right for the material but it will be turned down by most streaming services.

It will only be turned down when it is part of a playlist. Individual songs will often play at the higher level. However, the exact policies seem to be changing regularly so things could change next week.

The number one priority should be to get a mix that you are totally happy with. Don't even think about mastering, let alone releasing something, until you are happy with the mix.

Thank you James. I have some awareness of the general process/workflow - just rather limited experience in putting it into practice - and so this is very helpful. All the best.

[Zukan]

I mix in two ways: the old-school intern's method of listening to the mix at conversation level, and if something is lost, go back to the mix. I also listen loud, which allows me to hear detail that would otherwise be masked at low volume.

You can make anything loud whilst maintaining audio fidelity. Now, square wave compressed and bludgeoned to 2dB dynamic range...that's a no-no.

I use LUFS meters not for meeting streaming standard targets, but because of the amount of information provided, which, for a mix engineer/producer/come goat herder, is a lifesaver.