4.2 Audio Recordings Way Too Low Volume

Audio recordings are extremely low. I appreciate the effort of the new Normalize functionality, but I just want a raw, untouched recording of my set. With the introduction of Normalize, even without using it, audio levels are way too low.

All of these tracks hit and were at Level 10 audio. I expect that the recording matches this, having no CLIPS/PEAKS, but being very close to 0dB or -1dB. Since, you know, that’s what the set was. It certainly wasn’t played at -9dB, that’s rediculous.

None of these are even hitting -6dB, let alone -3dB.

This doesn’t make any sense and has really upset me as a regression of the quality of the Prime 4+ and Engine DJ OS.

1 Like

To be honest I was expecting this feedback on the recording level. Engine 4.0.0 introduced a significant 8dB boost in the recording level, with a limiter on top of it. The reason was that multiple people complained about the recording level being too quiet. They now removed that boost, and replaced it by an optional normalizing function. It is now called maximize, although it doesn’t limit. It just normalizes the highest peak in a recording to -1dBFS.

To be sure, this is peak normalizing, there is no loudness normalization of some sort. I’m just emphasizing this because you found the levels too low without using the normalizing. But since it is peak normalizing, it will almost always boost the recording’s levels. In your case you could expect an 8dB boost or so, assuming that your highest peak was on -9dBFS (judging from your screenshot).

Personally I would never want my recording to be limited, because if it is squashed too much, it cannot be undone. Also, the boost in 4.0.0 was only in the recording, not the master or booth outputs. This way it was really hard to judge how the recording would eventually sound, potentially leading to squashed recordings with the user only realize this after the recording was made.

For that reason, personally I am really glad with this change, because we now have optimal freedom to do anything we want in post processing. It is non-destructive, as opposed to an already limited recording. However we could see it coming that by undoing the 8dB boost, there would be remarks about the recording levels being too quiet again.

In my opinion a user preference would be the ultimate solution. We should have a recording level knob so that we could leave the recording level as it is, or boost it until it gets limited to have some more competent levels. This way, everyone would be happy. By the way, 0dBFS in the recording equals +10 on the meters Just so you know :slight_smile: . Edit: since 4.2.0, 0dBFS now equals +18dB on the meters, so the console’s outputs will clip before the recording itself will.

In your situation, the good news is that you can still normalize the recording, which is a very transparent process so it won’t sacrifice fidelity. Or you could even put a limiter on it, if you want more loudness without resulting in clipping.

5 Likes

Thanks for your thoughtful response.

My intent or desire is to simply have an accurate recording. If Denon wants to implement features and functionalities beyond that, that’s great, but I just want what I play to match what is recorded, that be the default option, and not have to sit and wait for any processing when saving out my set to file.

If I play a set and measure the volume coming out of the unit, I expect that to be the same volume being recorded internally by the unit. For example, if I was to use Audacity to record my set, it would come out as expected from the Main Output from the Prime 4+, but for some reason when recording using the Prime 4+ itself, the levels do not match at all.

I purchased the Prime 4+ for the standalone functionality. I wish to use the device itself to record and maintain the ability to stay standalone, and not rely on other hardware or software to do a pretty basic task of recording my audio as-is, as-heard and as-played.

When I play a set, my Main Output knob is set to 0. When I play sets, I ensure that the tracks hit 10 on the meter, but never clip. I check Main Output levels and they are exactly where I expect which is -3dB to -1dB, but the recording is at -9dB. What am I missing here?

There is already an error at this level,

Normally you should adjust your gains so that you reach 0 db on the vu meter of each of your tracks while avoiding exceeding it (first blue led on prime 4+ and first white led on prime4). Anything above 0db should be considered as headroom or as a safety margin.

And it is your final master level that you must increase or adjust accordingly. These are the basic principles of sound adjustment.

Even if I agree that this will certainly not solve your problem of recording too low, but playing your tracks at the limit of the last red led of the channel vu meters is a bad habit because you have very little margin before clipping

2 Likes

Currently, the recording actually is matched with what we play. Basically, when you clip the console it is safe to assume the recording will also be clipped. If you record hot enough, the audio should definitely have higher peaks than in the screenshot you showed. It would be potentially safer to not recording too close to clipping, because a small peak could ruin an otherwise good mix.

From a standpoint of a best practice, I agree. However, I tend to raise the signal a bit louder myself. In the end, as long as there is no unintended clipping, it should be fine.

Again, I really appreciate you sharing your knowledge and helping me to understand a bit better. It is very helpful in getting better and doing this properly.

What confuses me though is, if I have a properly mastered track, one that peaks at 0dB or -1dB, and I play it with my fader all the way up, with the trim perfectly centered, and the Master Output at 0…

Why doesn’t the track play at 0dB/-1dB and get recorded as such?

ALPHA21 - Into the Reality (Redspace Remix)

Sure, if the recording wasn’t mastered this way and was lower, I’d have to use my trim to get the audio level correct, but that’s the whole point of having a trim, eh?

I know where you’re coming from meLon, but basically Denon have set the recording level low as a safety feature, so that when DJs are mixing maybe four tracks at once, and not watching their levels, things don’t get distorted.

I’m sure we’ve all seen videos of DJs slamming their mixers into the red, right?

So the way Denon look at is it, rather than offer a recording level meter and recording level control so the users can set their own level, they just set the level really low, so that even if you go nuts with the gains and the EQs, there’s enough overhead that it shouldn’t be an issue.

It’s been discussed on here many times. They did raise the level slightly not long ago, after much requesting - but then others moaned that it was too high (even though there was still plenty of overhead) so what we’ve got now is a way to raise the level after recording. It doesn’t take long and it will bring the levels up to where you’d expect.

Yeah, I mix four tracks sometimes, typically Techno, and it’s a lot of fun.

I just make sure my tracks are at proper levels BEFORE ever bringing something into my Collection. I have a few tracks here and there I have from records that I have to pull the trim up on. This is especially important for newer artists, or people self-releasing tracks who don’t have the support of a team or someone to help them master.

I just don’t understand why the default isn’t “as-is” and anything beyond that an opt-in feature. Record my audio levels as I do my mix, by default, as they are played/heard. Enable a “Record with Overhead” mode for the people who want their recordings to come out at -9dB instead of where they actually were while playing.

I don’t want to invalidate people who need help with their levels, but why are people who have their levels correct and watch their levels being impacted by those who can’t?

As mentioned, 0dBFS in the recording equals +10 on the meters. Ergo, clipping your console equals clipping your recording. So, for a track with a ceiling on exactly 0dBFS you should play it on +10 on your console, this will result in a recording where the ceiling is also 0dBFS. However it can by tricky to record so close to clipping, so it would be a better strategy to lower your gains and use the normalizing option, which scans for the highest peak in the recording and brings up its level proportionally until the peak is on -1dBFS.

“Clipping your console equals clipping your recording”.

My whole complaint is that this isn’t true, since it’s dragging down the audio levels to -9dB for recording, which means there is a difference between what I played and what is recorded.

By default, there should be no difference in what is played and what is recorded.

By default, the unit should record what is played. Anything more should be a feature on top of this, but the ability to record what is played should be the priority and ensure no regression on that basic, core functionality.

Imagine you buy any other music equipment, sampler, sequencer, whatever, and you’re listening to what you’ve created an you like it, but when you export it it brings the levels down to -9dB and when you ask about it you’re told “Well, some people did this thing and so we had to do that”. Well, create a feature for those people, but it doesn’t make any sense for a professional grade piece of equipment to not output, by default, what the artist is creating. Its great to not get stuck in old ways, and to move forward and make cool new features, new workflows, and progress technology, but some software in the background messing with how I do things doesn’t sit right with me.

  • Record correct audio levels
  • Create a “Record with Overhead” option for those who need it
  • Create a “Normalize / Maximize” option for those who need it.

As I’ve explained in that other thread before, an 8dB boost is substantially and not at all slightly. It has a huge effect on workflow. I also explained that my gainstaging was totally fine as it was, my recordings never clipped. The 8dB boost also had a big effect when a user records way further from the ceiling, namely an 8dB level difference which is just huge.

A big downside of that boost was that it only affected the recording, so there was a huge risk the users found out that a recording was messed up only after the fact. There was no good way to tell how close a user was to the 0dBFS limit because it equaled a +2 value on the meters. But the meters go from 0 to +6 (at least on my Prime 4). It was imo a bad implemented solution for still a well existing problem.

As mentioned, we could have seen the remarks coming about the recording being too quiet now. Even a recording with peaks normalized to -1dBFS will not sound as loud as one where a limiter is applied. The problem with that is that when squashed too much, it can’t be undone. This is why I personally prefer to bring up the levels in post processing. On the other hand I can imagine when users prefer not to take that effort and just want competent levels in their recordings. I think the best option is a user preference, as mentioned :slight_smile:

I stand corrected. I just loaded in an audiofile with a sine wave and looped it. Then I turned up the gain until the meters overloaded. Then I dialed it back just until it wasn’t clipping anymore. I recorded this twice, one with and one without the normalizing applied. This is the not normalized recording:

And this is the normalized one:

So, it turns out that you were completely right that the ceiling is way lower than expected in the not normalized recording.

That is quite odd. Maybe the idea is that even though a dj is clipping while performing, the recording is not affected? I would have to test this further. But anyway, as mentioned I stand corrected, and this explains the low levels in your recording.

1 Like

As I also said in that other thread, 8dB (looking at it just as a figure on its own) is a lot of boost - BUT the initial record level was set so low, that even an 8dB boost was still way below the ceiling.

It was around 50% of the available headroom, so still should not cause any issues for people who are watching their meters and setting things appropriately.

2 Likes

I actually agree with Melon on this point about Denon letting the user of the controller set their own recording settings and limits. It would be easier for everyone involved.

3 Likes

I just did a bit more testing. Again I played a sine that was just below clipping. Then, I instant doubled it to another channel with equal settings, which means a 6dB boost in practice (L = 20 * log10(2) for the nerds). The signal is clearly boosted by that amount but still 2dB below clipping:

This means indeed that even when the console starts clipping, there is still 8dB of headroom in the recording. That is actually pretty nice!

1 Like

That would be the case when the levels don’t really exceed the 0dB point on the meters. But not everyone keeps the levels so low, which there is nothing wrong with, by the way. The point is that even when playing in the linear area (eg no clipping), it could still result in a squashed recording. Anyway, we’re in a quite different situation now. It is actually the other way around: even when the dj is redlining, there is still a chance that the recording isn’t clipped yet because of the 8dB extra headroom beyond the console starting to clip.

Anyway, I think it is clear that there are different preferences, which could be resolved by giving us some more control of the recording behavior :slight_smile:

In fact I will complete my previous post here is how any DJ should adjust their levels

  • set the maximum volume point to 0db on the channel VU meters for each track (the trim adjustment must be done in the loudest parts of the tracks)

  • Set the Master level so that the sum of all the mixed tracks (whether 2 or 4) reaches 0db on the master

  • Finally adjust the final volume on the amplifier/speaker/monitoring/or sound system to obtain the desired volume.

Why do this?

  • you keep a headroom on each channel that prevents you from reaching clipping even if you increase your EQ to make corrections.

  • You keep a safety margin on your master, which will avoid any distortion, or to engage the limiter which will flatten your sound, you keep dynamics and a better sound quality.

In addition, you avoid overheating the audio circuit and components (DSP, etc.) unnecessarily, you extend their lifespan.

  • You do not risk damaging or damaging your speakers or your sound system by subjecting them to too high an input load. It is the role of your amplified speakers, your amps or your sound system to amplify the sound, not that of your DJ equipment.

Your DJ equipment takes care of the mixing, your sound system increases the volume, so it is on this one that you must increase or adjust the desired final volume.

To make a point of comparison, know that when we produce a song, it is often produced at a very low volume during the composition/production stage. Often around -9 / -10 db.

This allows the sound engineer to keep a margin to be able to make all the necessary corrections during the mixing and mastering stage.

The final volume of the track is raised at the end during the mastering stage to reach the standards that we know today.

I think that Denon uses the same operating principle, recording very low to allow everyone to make all the corrections and post-processing they want on the mix before, for example, publishing it.

2 Likes

Those are definitely great guidelines, Gaian. However, in my experience this is how it goes in practice, whenever playing in the club or just with friends at home. The first dj starts playing, the levels are basically as you described. Then the next dj starts, he increases the gains a bit. The dj after that does this too. Etcetera. In the end, the gains are wide open because everyone wanted to be louder. I’m not saying this is the way to go, but this is just how it goes very often.

My signals are usually hotter than you described, but I never experienced any unintended clipping in my recordings. Do you still think my gainstaging is wrong?

Yes, speakers can be damaged when overloading them. But we could argue that that risk is bigger when there is more headroom. If the console is near clipping, it can’t be raised that much until the max is reached. This potentially decreases the chance of the speakers getting overloaded.

Anyway, I’m not arguing that you in fact bring up some very fine guidelines. But there is a nuance to all that. It is also a bit arbitrary since loudness between tracks can differ substantially.

You have a good point here too, but please allow me to add the following. I’m doing mastering for some artists and labels and often the premasters I get sent in are “-6dB” because “it is required to leave some headroom”. But this is mostly a myth. In some cases, a track slams a limiter in a daw, and producers will just lower the master fader before bouncing the pre-master. Now, the ceiling is 6dB lower but it is still squashed as much as before. The key is that the recording shouldn’t clip unintentionally. If some masterbus processing contributes to a sound the producers wants, there is nothing wrong with that too. The producer has to realize though how it would translate in the end product. Other than that, as long as the recording is not clipping it should be fine. It doesn’t matter in that case whether it peaks close to 0dBFS or at -6dBFS, or any other value. In fact, with floating point audio there is even headroom above 0dBFS, so technically a pre-master can exceed 0dBFS and will still be fine to work with.

Having said that, I am indeed happy to be able record without any non-linear processing (either clipping or limiting) applied to the recording itself, to have all freedom in post-processing. So the point you wanted to make is still very valid.

2 Likes

I totally agree with you, and I also see this every time I play with other DJs, especially in clubs.

The problem is that we are dealing here with a lack of knowledge of 90% of DJs of the basic rules of sound adjustment.

Having fun mixing two tracks with automatic sync and turning the knobs to look cool like James Hype is within everyone’s reach.

Having basic notions of sound adjustment and processing is not as much, and for good reason, it is definitely less fun than seeing all the LEDs of the vu-meters light up like a Christmas tree.

What I am telling you here is however the basics that any DJ should know first, even before learning to beatmatch. It is as basic as learning to turn a steering wheel when you want to learn to drive, but unfortunately 90% prefer to skip this step.

So a little reminder on the best practices for adjusting the sound and on the role of each piece of equipment in the chain doesn’t hurt. The role of the DJ system is to mix, the role of the sound system is to amplify, not the other way around.

4 Likes