Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Audionut11]

If I'm watching a movie, and decide that the loud bits are getting to loud for the neighbors, DRC provides the fix by reducing the level of the peaks, while also increasing the level on the dips.

Applying DRC, and then having to turn the volume level down because everything except the peaks have been boosted, doesn't make sense.  It is the peaks that are disturbing the neighbors, so boosting everything towards peak levels is counter intuitive.  I should be able to apply DRC to not annoy the neighbors, without also having to reduce the volume level.  Then at some stage after this, disable DRC, and have to change the volume level again.

I'm pretty sure that DRC was originally not intended to adjust the loudness (or average volume or whatever you want to call it).

Now lets look at it from a SNR point of view.

When you apply digital boost, you boost the noise by the exact same level as you boost the signal.  It should be obvious why you don't want to apply large levels of digital boost.
When you reduce level, you have a negative effect on the transient response of the signal.  It should also be obvious why this is an advantage in a DRC situation.

An excellent feature would be one similar to that applied by Yamaha, they call it adaptive DRC.  The level of DRC is adaptive based on the volume level.  This is an excellent feature, since it provides varying levels of DRC (including no DRC) based on the volume level without further user intervention.  Generally, as you reduce volume level, you actually want some boost applied to help reduce the chance of the signal reducing below the noise floor of the environment, and, you want some reduction of the peaks, since this is probably why you were reducing the level in the first place, to reduce the level of the peaks.  When volume level is +/- 0dB or greater, no DRC is applied.


edited for spelling.

[JimH]

Try MC'S Night Mode.

[wer]

Unfortunately Night Mode in DSP Studio makes everything louder (even the loud bits) forcing you to turn down the volume, just like Audionut was trying to avoid.

[Audionut11]

My post was cut from the release thread, where I was sharing my thoughts about the discussion that was happening there:  https://yabb.jriver.com/interact/index.php?topic=100902.msg700859#msg700859

Try MC'S Night Mode.

In my humble opinion Jim, night mode is counter intuitive (I would go as far as to say broken), as per my description above.

[wer]

I think it's awesome that JRiver is trying to implement Dynamic Range Compression, but I agree that Night Mode doesn't work the way it should or would be most beneficial.  Jim and his team are generally pretty great about making improvements though, so I'm hoping this will be one of them.

[JimH]

The build thread is not a good place for discussion.

[wer]

Ok.  Jim, would you please weigh in on whether Night Mode needs adjustment, or is working as designed?  Currently it makes peaks louder as well as quiet parts, so it's not a good solution for the nighttime listen problem.

I think that dynamic range compression that better preserves average volume level by both cutting peaks and boosting valleys better serves the pursuit of volume leveling for videos by making it less necessary to adjust the volume when you turn night mode on or off.

Another way of looking at, you're trying not to disturb others watching a movie, so you've set the volume at a level where you can hear the dialog (perhaps barely).  Then music/explosion/whatever comes on and it's "Wow that's too loud!"  You should be able to turn on Night Mode, problem solved.  But that's not what we currently have: Night Mode boosts the explosions which is the exact opposite.

Could you address the question of the algorithm's design and whether boosting both peaks and valleys in the pursuit of compressing dynamic range is the correct and appropriate thing for Night Mode to be doing? 

Thanks...

[blgentry]

Currently it makes peaks louder as well as quiet parts, so it's not a good solution for the nighttime listen problem.

That's not what's happening as far as I can tell.  Night mode boosts low level sounds and leaves the peaks alone.  That's what it's supposed to do.  If it boosted both, that would just be the same as turning up the volume, which isn't useful.

I think that dynamic range compression that better preserves average volume level by both cutting peaks and boosting valleys

I don't know of any Dynamic Range Compression system that does this.  It's one or the other, not both. 

Another way of looking at, you're trying not to disturb others watching a movie, so you've set the volume at a level where you can hear the dialog (perhaps barely).  Then music/explosion/whatever comes on and it's "Wow that's too loud!"  You should be able to turn on Night Mode, problem solved.

I think that's a very fair and accurate way of describing the behavior of a DRC that lowers the peaks.  Much like a hardware compressor that you would use in a live sound application, or a studio.  If JRiver decides this belongs in MC, it should be an option, as opposed to the only "night mode".  Why?  Because there is LOT of discussion here about the Volume Leveling tools and how they lower overall volume.  Some people REALLY don't want to see the overall volume lowered any more than it must be to achieve the task.  So the current night mode really should be available for those people.

I'll make my alternative suggestion again:  Use the current Night Mode *and* set up an overall gain reduction of 10 to 12 dB as one of the Parametric EQ settings.  When you click the box for Night Mode, click the box for the EQ too.  I'm pretty sure this would solve your problem.

Brian.

[Audionut11]

I don't know of any Dynamic Range Compression system that does this.  It's one or the other, not both. 

http://www.dolby.com/us/en/technologies/all-about-audio-metadata.pdf

When dynamic range profiles are asserted within the decoder, the decoder raises the level
of the softer portions of the program while lowering the level of the louder portions,
allowing the user to enjoy the movie without having to continually reach for the volume
control.

And before this section it actually says.

Line Mode provides a lighter type of compression, and also allows user adjustment of the
low-level boost and high-level cut parameters within a home decoder. This adjustment or
“scaling” of the boost and cut areas allows the consumer to customize the audio
reproduction for their specific listening environment.

That's a feature I was unaware of.

Here is some Yamaha literature of their Adaptive DRC.


Looking at the right hand image (volume control high), we can see an equal amount of boost and cut.  Here, average volume level is not affected.
With the left hand image (volume control low), we can see larger amounts of boost then cut.  This makes sense in this situation since it's based on the volume control, where a lowering of the volume level (by definition) lowers the peaks.

The handy thing about the above implementation is that it doesn't require any user intervention.  When you turn the volume down late at night (or whenever), you automatically gain the boost in low level (compression) to offset the effect of the volume control.

[wer]

That's not what's happening as far as I can tell.  Night mode boosts low level sounds and leaves the peaks alone.  That's what it's supposed to do.  If it boosted both, that would just be the same as turning up the volume, which isn't useful.

Unfortunately, that's exactly what it's doing.  I've checked with the analyzer in DSP Studio (to validate my ears).  It makes peaks louder too.

[BryanC]

Unfortunately, that's exactly what it's doing.  I've checked with the analyzer in DSP Studio (to validate my ears).  It makes peaks louder too.

Have you tried enabling R128 leveling as well?

[wer]

Volume leveling is enabled.

I don't see an option to enable/disable R128 separately from anything else, so I didn't know that was possible; I understood it was baked in to MC21.  How do I do that?

[blgentry]

Unfortunately, that's exactly what it's doing.  I've checked with the analyzer in DSP Studio (to validate my ears).  It makes peaks louder too.

If that's true, I would consider that to be broken.  The only way I know to test this is to find the loudest part of the movie and then play it with and without night mode and see what the levels are.  Doing it with a very short movie would probably be the easiest test.

I'd enjoy hearing from someone who really knows how Night Mode is supposed to work.  I'm making assumptions all over the place here based on what I know about audio.  How is Night Mode supposed to work?

Brian.

[BryanC]

If that's true, I would consider that to be broken.  The only way I know to test this is to find the loudest part of the movie and then play it with and without night mode and see what the levels are.  Doing it with a very short movie would probably be the easiest test.

I'd enjoy hearing from someone who really knows how Night Mode is supposed to work.  I'm making assumptions all over the place here based on what I know about audio.  How is Night Mode supposed to work?

Brian.

From the feature description and based on how similar DSPs have acted in the past, it should only boost the volume of the quiet portions of the movie (thereby decreasing the dynamic range).

Relevant discussion here: https://yabb.jriver.com/interact/index.php?topic=82025.0

[mwillems]

Given that most recorded audio (both for video and otherwise) is mastered so that the highest peak is at or near 0dBFS, there's no way that MC could boost the highest peaks without pushing everything into clipping.  The night mode compression raises the average level of the recording by boosting almost all the audio to some extent, quieter portions are boosted significantly more than the louder portions, but the very loudest portions literally can't be boosted.  Think of a layer cake between two pieces of glass: if you raise the bottom pane of glass while keeping the top pane stationary the cake will compress and the layers (other than topmost layer) will all rise to some extent.  The bottom layer will rise the most (as a percentage), and while the bottom of the topmost layer will rise a little, the top of the topmost layer will not rise at all (because it can't).

All of the adaptive volume settings boost the volume, just in different ways (some don't apply compression, some do).  Night mode does make things (on average) momentarily louder (especially if you were using volume leveling ), but given that the setting is three menus deep, I'm not really sure I understand the "pressing night mode in panic to avoid disturbing the neighbors," when the volume and/or pause button are much easier to reach and are likely to be the first reaction.  i agree it would be nice if night mode played better with volume leveling, but this seems pretty easy to work around?

[gvanbrunt]

I just wanted to point out that when reducing the dynamic range to avoid peaks you raise the volume of the quieter parts. This makes it louder overall. Cutting down only the peaks would result in a worse sound to noise ratio. So you do have to use the volume control and turn it down. Perhaps an improvement could be an automatic reduction in volume that you can set as an option.

[Audionut11]

Given that most recorded audio (both for video and otherwise) is mastered so that the highest peak is at or near 0dBFS, there's no way that MC could boost the highest peaks without pushing everything into clipping.

What about when the volume level is below 0dBFS?  With the volume level at -10dBFS, there is 10dB of headroom to increase the volume of the peaks.

Cutting down only the peaks would result in a worse sound to noise ratio.

Cutting down only the peaks would be daft.  Any reduction in volume, including when you reduce the volume control lowers SNR.  This doesn't just apply to reducing peaks.

When you apply boost to the lower level portions of the film, the noise doesn't magically stay at -80dB, or whatever level is in the program material.  When you apply large amounts of boost to the lower level portions on the film, the noise gets boosted by the same amount.

There will always be a perceptual increase in volume when compressing the content (no matter how you compress it).  Currently, there is a large increase in volume.  IMHO, a better balance could be found if the developers are interested in spending the development time on it.

[gvanbrunt]

Sorry, but you mistaken on several points.

First, mwillems is correct that most tracks are mastered so the peaks are close 0dbfs. So if you have something in which the peaks are at -10 either it is mastered wrong, on something else is going on with your testing or setup. The track itself is going to be close to 0dbfs. What I think you are doing is mixing up the volume with the levels on the track itself. MC works on the source, and does not try to manipulate volume levels in some way. Trying to mess with the volume somehow would be extremely problematic. To put it very simply the changes happen in the signal chain long before anything relating to volume are applied.

Secondly, you are assuming the noise only comes from track itself witch is not correct. Noise is introduced all over the place and "reducing the signal strength" of the source would allow that. Again I believe you are mistakenly thinking that the changes are made by manipulating the volume itself further down the chain, which is not what happens.

[blgentry]

From the feature description and based on how similar DSPs have acted in the past, it should only boost the volume of the quiet portions of the movie (thereby decreasing the dynamic range).

Relevant discussion here: https://yabb.jriver.com/interact/index.php?topic=82025.0

I searched that thread for "night" and found no references to night mode at all.

Here's what I'd like to know:

1.  What is the threshold for night mode increasing the level of low level sounds?  This should be measurable and should be part of the algorithm.
2.  What is the amount of gain applied?

These are fairly standard Compression parameters.  Assuming that Night Mode uses something like a "normal" compression algorithm.

I think these are fairly reasonable and easy to answer questions.  Pretty please. 

Brian.

[Audionut11]

Secondly, you are assuming the noise only comes from track itself witch is not correct. Noise is introduced all over the place and "reducing the signal strength" of the source would allow that. Again I believe you are mistakenly thinking that the changes are made by manipulating the volume itself further down the chain, which is not what happens.

You're trolling me right!

The listening environment contains noise.  Increasing the volume (making the noisy bits make more noise) raises the SNR of the system as a whole, until such point as the noise in the source, or the noise generated by all of the electronic bits is greater then the noise in the environment.

Since a significant proportion of the content available today, and a significant proportion of the systems in use today, have a noise floor below that of the significant proportion of listening environments, any reduction in volume (doesn't matter how you are manipulating the peak level of the signal), by definition, reduces the SNR of the system as a whole.

Please stop wasting time attempting to assume what I am assuming.  If you need clarity, just ask.

So if you have something in which the peaks are at -10 either it is mastered wrong, on something else is going on with your testing or setup.

Or..............the volume control has been reduced 10dB, which by definition, raises the headroom of the system.  The point being, and I'll make myself clear so you don't have to assume, I don't care what level the source was mastered at, there is headroom available in the system for the volume to be increased without clipping from the simple act of reducing the volume level.  It is this fact that allowed the implementation of this:  http://yabb.jriver.com/interact/index.php?topic=99204.0

I don't know the code, so I can't comment how it works.  If the code isn't supposed to work this way, but does, then clearly it's a bug.  I'm happy to be told that it does not work this way, and completely disregard this angle of reasoning.  But I'll need more then assumptions about what I appear to be assuming.

Thanks.

[gvanbrunt]

Audionut11,

I think you need to tone it down a bit. People here are trying to help you and having an attitude is not going to help that. Perhaps you are used to other forums where this is allowed. It is not here so don't stray too far down that path or an Admin will likely "give you a timeout".

Given your tone I'm not prepared to explain this further but perhaps someone else is.

[Audionut11]

You're making assumptions again.

http://www.catb.org/esr/faqs/smart-questions.html#not_losing

Community standards do not maintain themselves: They're maintained by people actively applying them, visibly, in public. Don't whine that all criticism should have been conveyed via private e-mail: That's not how it works. Nor is it useful to insist you've been personally insulted when someone comments that one of your claims was wrong, or that his views differ. Those are loser attitudes.....................................Exaggeratedly “friendly” (in that fashion) or useful: Pick one.

[mwillems]

You're making assumptions again.

http://www.catb.org/esr/faqs/smart-questions.html#not_losing

"Community standards do not maintain themselves: They're maintained by people actively applying them, visibly, in public. Don't whine that all criticism should have been conveyed via private e-mail: That's not how it works. Nor is it useful to insist you've been personally insulted when someone comments that one of your claims was wrong, or that his views differ. Those are loser attitudes.....................................Exaggeratedly “friendly” (in that fashion) or useful: Pick one."

You're trolling me right!

Accusing another forum member of trolling when they're trying to answer you in good faith, and then indirectly telling them they have a "loser attitude" is impolite, and requires making no assumptions (your tone is aggressive and unhelpful).  You can disagree with someone and still be polite.

You're unlikely to get good answers from anyone here that way, certainly not from me.  I prefer "friendly and useful." The "or" in your quote is presented out of context and creates a false dichotomy.  Eric S. Raymond's article is guidance for the person seeking information, i.e. you, not for the people spending their free time trying to help you.  

[Audionut11]

Sigh!

Accusing another forum member of trolling

https://en.wikipedia.org/wiki/Exclamation_mark

A sentence ending in an exclamation mark may be an exclamation (such as "Wow!", "Boo!"), or an imperative ("Stop!"), or may indicate astonishment:

http://www.thefreedictionary.com/astonishment

Great surprise or amazement.

I specifically added the exclamation mark there.  Clearly, the statement can be construed in various manners, and on this occasion, was construed in the wrong manner.  I specifically stated above that if clarity was needed, it could be asked for.

Since the manner in which I reply is open to interpretation, and is clearly being interpreted in the wrong manner, I shall take my leave so that this thread can return to one of usefulness.

Cheers.