Are you REALLY listening at Reference Volume??? Maybe Not...

[JStewart]

first, I’m calibrated to -20dB pink = 79 dBC when my volume is at 0dB ie reference. I’m then measuring Dialog on Blu-ray titles and typically finding that for many movies my volume control has to be around -6 or so to give me 71dBC Dialog.

So then, conversely, the dialog level would have been 77dB had the master volume not been lowered? And that after calibrating to 79dB. So what then, if calibrated to 85dB the dialog level would have been 83dB?

If I’m looking at the numbers properly, I wonder why the dialog would have been that loud to begin with? Could it be entirely related to psychoacoustics relative to room size?

Anyway, the idea of calibrating to the dialog level (the anchor element?) with the rest of the mix being relative to it, makes total sense in my mind and works for my ears/brain combo.

Thanks again for the great idea!

 

[Paul Hurt]

I’m checking back thru my figures and yes, good point, I might be a bit off somewhere. But not by a ton. When I was calibrated to 85 and running at reference, average dialog levels from blurays typically sat in the high 70s to low 80s dBC. Quite a big range.

Interestingly, DVDs pretty much all had Dialog in the low 70s.

I think either post-production folk are either mixing/mastering blurays too loud, or potentially setting dialNorm wrong… or both.

But as you say, the best approach (if people can be bothered!), which takes all the calibration and reference issues out of the equation, is to grab your disc or streaming content, find a scene with some ordinary dialog and measure the rough average SPL of the speech (ignore the silence between words!), then work out where your master volume needs to be to shift that level to 71dBC (or whatever you prefer - 68, 69, 70, 71, 72 perhaps), and leave your volume control at that setting for that movie. Maybe write it down for next time too

Bit of a song and dance obviously, but it works.

 

[VJM]

@Paul Hurt how did you come up with -6 as reference, according the chart here it depends on the cubic feet of the room, mine is in 1,600 cubic feet so -7 is where I'm at, I kind of like that because the Dolby cinema processors reference is 7

 

[ddude003]

Sounds like there is a relation like sound roll-off law of inverse square plus some fudge factor for home theater systems...

As far as post production mixing and mastering too hot goes... Once you over cook a steak you can't un-cook it...

 

[Paul Hurt]

Hey VJ - was good to (virtually) chat with you and everyone tonight.

My suggestion to think of true reference as being 6dB lower than we used to set it was because most small post-production facilities I heard about (and people building mix rooms at home) were calibrating to 79dB. Skywalker Sound meanwhile were calibrating to 80. So even tho ATSC were recommending 78, in practice almost everyone was saying they were using 79dB.

But obviously it’s to taste. Go louder if you’re happier a bit louder (like Todd for instance). Or go quieter if that suits you better. -7. Why not.

Bearing in mind of course that it all depends on the title you’re watching anyway. Most blurays appear to still be too loud at -6. If I want dialog to sit just over 70dB for Avengers:IW and Avengers:Endgame, my volume needs to be 5 to 6dB lower than that! (So volume at around -12 in your money). Watching Judas And The Black Messiah on Amazon Prime, it needed to be 2dB *over* my new reference (-4 to you) to get that same dialog level.

Did you also notice that Jon in the chat tonight was saying that when they’re organising a movie premiere, they sit in the theater and turn the playback level up and down until they find the point where they feel it’s about right!? Seems even 85dB-calibrated big theater systems cant guarantee an identical experience from one venue to the next. Although I’d bet they start at reference level first, then adjust to taste from there.

One thing I have noticed after all of this is that I’m now very aware of the difference a change of 1dB makes to a soundtrack.

 

[Paul Hurt]

As far as post production mixing and mastering too hot goes... Once you over cook a steak you can't un-cook it...

Perfect analogy. You can’t uncompress a mix that’s already got too much compression going on in it.

Music mixers in particular sometimes don’t realise that they’re using compression as a crutch, and not because the music is better for it. Compressing instruments often makes them easier to position in a mix, but at the end of the day, quite often it’s better to just move the fader if you need to control the level of something throughout a song.

 

[ddude003]

Sometimes you just have to trust your ears... When I watch movies, streaming or BlueRay, I turn on my AVR which defaults to power up at -50db... The AVR does center, side and behind speakers... Center is +3db above sides and behinds... I play for a bit and adjust center to taste with my eyes closed... Then I match/blend my Left and Right until it sounds good... I pay no attention to the numbers... Then I start the movie from the beginning, open my eyes, and enjoy... Sometimes you just have to trust your ears... Old School... Oh, yea, Left and Right and Sub are thru a Tubed Pre and McIntosh Amp...

Not just compression, often just run in too hot (gain) not leaving any headroom...

 

[VJM]

I found something that might interest you, on AVS there's a thread for BEQ (Bass EQ). For now, ignoring the Bass EQ part, part of their process is to show a master volume (MV) offset from reference + or -. They analyze the raw audio file and can see the average levels and if it's off from reference (85db) and they have a recommended Master Volume adjustment on your AVR to listen at, here's a few examples.

Jaws 4K - Atmos

Tenet 4k - DTS-HD MA 5.1

The Tomorrow War 4k Stream - DD+ Atmos

The Empire Strikes Back 4k - Atmos

For example, if your normally listening level is at -6 (your preference reference) then watching The Tomorrow War this would be offset by -5.5 so you would adjust to -11.5, Empire Strikes Back is +1.5 so your personal reference offset would be set to -4.5.

We'll have to test this out and see if this checks out, of course there is the Dolby DialNorn difference also on Dolby tracks and BEQ started digging into that recently, but we'll save that for later

 

[Paul Hurt]

Thanks, interesting, I’ll look at that. It’s very similar to what I’m doing… tho I’m gearing my method towards achieving consistent dialog level… trying to hit a target average level across the whole movie will skew things depending on the distribution of loud segments to quiet segments (which is partly what I was getting into in the podcast with the discussion of broadcast and the target programme LUFS spec). The idea with movie mixing is that loud movies should stay loud, and predominantly quiet movies should remain quiet IMO.

Although sticking an entire soundtrack thru a broadcast LUFS meter like the TC Electronic radar thing (can’t remember what it’s called) would be a handy if “broadcast-style” way to level-match movies. None of this is an exact science anyway. It’s ok to be a dB or two off.

Our figures may match up quite well anyway. I think it’s a sign of progress that we’re finally trying to get a hold of this

 

[VJM]

Just noticed this but the interesting thing is if you look at Tenet and know how Christopher Nolan is with the sound of his movies. According to the data from the BEQ database Tenet volume should be reduced -6.5, right where the small listening room volume should be around. This makes me think that Nolan had them supply the theatrical mix on the disc, spot checking a couple other of his releases from BEQ..

Dunkirk MV offset -5
Interstellar MV offset -5.5
The Dark Knight Rises MV offset -5
The Dark Knight MV offset -4

Very interesting, we'll need to check this out and report back, I will soon as I can.

 

[Paul Hurt]

Great observation, another thing I’ll check out. It keeps getting more interesting, doesn’t it ?

 

[VJM]

Just noticed this but the interesting thing is if you look at Tenet and know how Christopher Nolan is with the sound of his movies. According to the data from the BEQ database Tenet volume should be reduced -6.5, right where the small listening room volume should be around. This makes me think that Nolan had them supply the theatrical mix on the disc, spot checking a couple other of his releases from BEQ..

Dunkirk MV offset -5
Interstellar MV offset -5.5
The Dark Knight Rises MV offset -5
The Dark Knight MV offset -4

Very interesting, we'll need to check this out and report back, I will soon as I can.

This seems to work very well, I'm working from home so I took a break and went into my theater and tested this out on random movies from my k-scape.

Tried Back to the Future Atmos MV offset of +3 so I set my volume to -4 from my default of -7
Inception DTS-HD MA - MV offset of -1.5, set my volume to -8.5 from my default of -7
Blade Runner Final Cut Atmos offset of +1 set my volume to -6 from my default of -7

Oh and here's the link to the BEQ database https://beqcatalogue.readthedocs.io/en/latest/

 

[VJM]

Great observation, another thing I’ll check out. It keeps getting more interesting, doesn’t it ?

For sure, but it's been fun

 

[Paul Hurt]

For sure, but it's been fun

It has. Plus I’m enjoying movies more. Means I’m not guessing at a volume setting then having keep the remote on hand and ride the volume the whole time. Now I have this methodology, movies are the right volume for me from the start.

Despite this, Judas & The Black Messiah got pretty loud towards the end… Quite a dynamic movie, big dialog swings, so it’s easy to set the vol too low and miss some dialog.

But I think another factor that affects your loudness perception is how late at night it is and how close your neighbours are

 

[VJM]

Not sure if you watched the streaming version of Judas & the Black Messiah but according to the BEQ database for the streaming version:'

Judas and the Black Messiah
DD+ Atmos
MV Adjustment: +7.5 dB
+1dB dialnorm compensation

This is mastered really low

 

[ddude003]

Wonder what the logic is behind having a Streaming version that is different than a Blue Ray or DVD version... Is it so it sounds good on TV or Sound bar type systems? Are they trying to save some bit space for transmission across the net?

 

[VJM]

Wonder what the logic is behind having a Streaming version that is different than a Blue Ray or DVD version... Is it so it sounds good on TV or Sound bar type systems? Are they trying to save some bit space for transmission across the net?

Yeah, lowest common denominator, like you said for a soundbar or TV, that's probably 90% of how it's played back

 

[Paul Hurt]

Wonder what the logic is behind having a Streaming version that is different than a Blue Ray or DVD version... Is it so it sounds good on TV or Sound bar type systems? Are they trying to save some bit space for transmission across the net?

If it’s different, I suspect it’s because most of the streamers work to R128-style LKFS/LUFS audio specs. Judas And The Black Messiah sounded fantastic on Amazon. It needed me to run at +2 (on a 79dB ref system) to get 71dB dialog. But believe me, the loud bits were still LOUD.

Note that the current broadcast specs leave space for just as much if not more dynamics than theatrical and blu-ray.

If anything, Judas ATBM was mastered about right on Amazon. It’s blu-rays that are usually too loud, which potentially reduces available headroom on the disc. Yes, sometimes it feels like a mix has been “made for soundbars etc”, but that can be apparent on any media, not exclusively streaming. Age Of Ultron on Blu-ray being the worst example that springs to mind.

We ought to get past this misconception that louder = better. It’s usually the opposite. Just like CD, it’s often the loudest ones that have the least dynamics.

If it feels quiet, just turn the volume up and you’re good.

 

[Paul Hurt]

I ought to point out also that Dolby Digital incorporates built-in dynamic range compression algorithms that kick in only if the programme is being played back on a device with limited power and dynamic range, like a soundbar or a TV internal speakers. Played back on a home theater receiver the DRC would not engage, so you’d get the full dynamic range of the original mix.

So really, there shouldn’t be any need to deliver “made for soundbar” mixes, because the decoder in the playback audio device is meant to automatically handle with that for the viewer.

It’s not clear to me yet how well (or if!) that’s working in practice mind you. You prob know can manually switch on Dolby’s dynamic range compression on a receiver too if you want. I’ve never liked the sound of it, but then I suppose I wouldn’t .

 

[JStewart]

Just noticed this but the interesting thing is if you look at Tenet and know how Christopher Nolan is with the sound of his movies. According to the data from the BEQ database Tenet volume should be reduced -6.5, right where the small listening room volume should be around. This makes me think that Nolan had them supply the theatrical mix on the disc, spot checking a couple other of his releases from BEQ..

Dunkirk MV offset -5
Interstellar MV offset -5.5
The Dark Knight Rises MV offset -5
The Dark Knight MV offset -4

Very interesting, we'll need to check this out and report back, I will soon as I can.

I checked out Interstellar DTS Master HD 5.1. BEQ guideline 5.5dB down from a dialed in setting I’ve been using was good from the standpoint of I didn’t want the loudest parts to be louder. It became uncomfortable when I tried turning it up during loud scenes.

I found the dialog level inconsistent. I felt during some loud scenes the dialog was louder because the characters appeared to be visually speaking at a conversational level but could still be easily distinguished above the loudness of the scene. Impossible to get meaningful data with an SPL meter that’s picking up all sound though. I guess if you have to speak while passing through a wormhole you’re going to have to speak a bit louder.

During some quiet scenes, where there was dialog only, the dialog level measured below 65dB while others were at 70dB.

I‘d conclude the BEQ level adjustment is a good guideline for this film because with it, most dialogue was at a level I’ve found works in my room. That said, for me this one is really hard to set the volume and leave it. Be interested to hear what anyone else’s thoughts are.

 

[VJM]

There's definitely usually a difference between the streaming mix and the disc, for example Top Gun: Maverick.

DD+ Atmos streaming version it's recommended MV offset of +7 and the disc version TrueHD Atmos there's no offset so should be good with MV set at -6 for reference.

I checked out Interstellar DTS Master HD 5.1. BEQ guideline 5.5dB down from a dialed in setting I’ve been using was good from the standpoint of I didn’t want the loudest parts to be louder. It became uncomfortable when I tried turning it up during loud scenes.

I found the dialog level inconsistent. I felt during some loud scenes the dialog was louder because the characters appeared to be visually speaking at a conversational level but could still be easily distinguished above the loudness of the scene. Impossible to get meaningful data with an SPL meter that’s picking up all sound though. I guess if you have to speak while passing through a wormhole you’re going to have to speak a bit louder.

During some quiet scenes, where there was dialog only, the dialog level measured below 65dB while others were at 70dB.

I‘d conclude the BEQ level adjustment is a good guideline for this film because with it, most dialogue was at a level I’ve found works in my room. That said, for me this one is really hard to set the volume and leave it. Be interested to hear what anyone else’s thoughts are.

Thanks for your thoughts on this, one thing about Interstellar is that there're a lot of complaints about the dialog being hard to understand so more then likely it's part of the mix and your description sounds similar.

'Interstellar' doesn't have sound problems — Christopher Nolan wanted it that way

Sound designer says it's about understanding emotion, not every word

www.theverge.com

 

[Paul Hurt]

Yep, can be hard to get a “read” on dialog levels if they jump about a lot. Judas ATBM was very much like this too. That’s why I’m measuring with SPL logging in REW. Having that graph makes it easier find both the loudest and the quietest bits of dialog in a speaking scene (ignoring any deliberate whispers or shouts) then take the average of them.

Obvs I try not to measure any dialog scenes with background noise - either the background will push the SPL up itself, or the characters can be speaking louder or be mixed louder, to get over the background. (Except in Nolan movies ;-)

It’s def not an exact science tho, for sure. Sometimes it’s just a best guess, and then I see how it goes. This method seems to get me close tho, and I prefer it to just starting at (new) reference level, and then finding it takes half the movie to establish where to leave the volume control.

 

[JStewart]

Yep, can be hard to get a “read” on dialog levels if they jump about a lot.

To demonstrate your point, here are some examples from Interstellar. For these measurements the volume was set to 72dB with -20dB pink noise on the center channel. A bit lower than what I've found is right for a typical film for my system in a ~1450 cubic foot room. As you will see, it varies quite a bit.

In this one you can see the change in dialog level between the front porch scene and the ball game scene at 1:25 into the log.

Here you can see the dialog change between the conference room scene and the tour that follows. That's like a 6dB change which perceptually pretty big.

This is Romilly explaining the wormhole to coop. Down about 5dB from the school conference scene (1st chart).

And just for fun, this starts at entering the wormhole. This was recorded -20dB from all the previous charts in this post. Hey, it was 7am here! This would have been 102dB at my setting and makes me question whether the absolute decibel values can are correct.

The LCeq 1 minute average appears a good way to judge volume level of a scene vs. eyeballing the logged C weighted slow SPL. What do you think?

I'm using REW V5.20.14

I'm on board with your methodology. While it can't be perfect due to variability of dialog within a mix it's better than setting to a (somewhat) arbitrary value.

It would definitely be easier to use the BEQ gain adjustment recommendations. Going forward, I'll continue to compare that vs settings based on dialog levels to see if there's a consistent correlation.

Be nice if there was software to do this! Dolby LM100 is available for short money.

Also, I've read that dialnorm values are "typically" -27dB from fullscale. That would equate to 78dB. In my room I've found -9dB or -10dB from reference is where I land after setting based on a dialog volume. It's interesting that this correlates to:
1) 75/76dB for reference (with -20dB pink noise) which is is the number in the chart in post 28 from @VJM
2) The "typical" dialnorm value.

So if one had the actual dialnorm value used in conjunction with the recommended reference SPL in post 28, would one presumably have the volume correctly for each individual film?

 

[Paul Hurt]

Yes, interesting. Think we’re all converging on similar conclusions.

Reading those graphs is challenging because obviously you need to indentify what’s dialog and what’s silence. Generally the peaks are Dialog, the valleys are gaps between lines/words.. so I’m taking the value of the highest peak and the lowest peak as the high and low Dialog values.

The release time of slow response is SO slow that the valleys are more a measure of how long a silence is between lines of speech! The longer No-one speaks, the lower the trough goes.

I’m aiming to grab no more than a minute of dialog for consistency. Too much and it’s too hard to read.

Its funny tho, because you beat me to the punch there - I was also wondering if it might be possible to measure an entire chunk of a dialog scene as Leq, to directly get a value for average Dialog level.

I need to figure out exactly how Leq works.

Does it equate that to what we’re doing? Maybe it’s easy to translate to our numbers.

Re dislnorm, it relates to the level the dialog is mixed at on the meters, rsther than the SPL in the room, so it’s a bit impossible to know if dialNorm has been set accurately by measuring SPL.

One of the things that *might* explain why blurays always seem a bit loud is that the encoder usually has dialNorm set to -27 as default, but the recommended target dialog level for broadcast is often quoted as -24LKFS. So it’s possible people are mixing Dialog louder than the dialNorm is set to, which would cause levels overall to be 3dB too high,

Could be wrong tho!

Fun.

Still lots to investigate for me, not enough time right now.

 

[Paul Hurt]

So, just to relate what I’m saying about guessing an average Dialog level, the min dialog SPL in that Judas ATBM chart is 68dBC, and the max is 74 or 75dBC. Average of those is 71dBC, which is what I’ve settled on as a good target dialog level in my room.

(That’s because that graph was measured AFTER I’d established my preferred listening level for the film… which was +2dB above new 79dBC reference)