Poll: Should Sony use DTS-HD Master Audio or Dolby TrueHD?

[PeterTHX]

Originally Posted by Monkey

I think we've been getting about 1 vote per day now for a week or so.

779 (85%) to 127 (15%) currently.

Uh huh, that's some optimistic math there.

[Anthony P]

Originally Posted by BIslander

Nonetheless, it seems clear that the DN value is fundamental to important Dolby processing functions that are not related to loud TV commercials. For example, how do you do proper DRC unless you know where the dialog is pegged in relation to full scale?

DN is important to DRC, I pointed out many pages ago that it was one of its use, but you don't really need it, in the end you need a 0 point (the point that does not move) where everything contracts or expands around. DN dictates that point based on the movie, but you could just as easily use any other point and DC around it.

[Anthony P]

Originally Posted by CraigW

DRC and DN are metadata meaning they only are instructions for the decoder.

yes but the decoder uses that metadat and alters the audio based on it.

DN is the only one that cannot be defeated and again IT DOES NOT ALTER THE PCM TRACK that is created during decoding.

but no matter how many times you state that, it is still wrong. It does alter the PCM, if it did not then why would it be there in the first place.

Like you pointed out with the link to Dolby’s site

In contrast, a louder program requires a shift to match the –31 dB standard. When the Dialogue Level parameter setting is –21, the decoder applies a 10 dB level shift to the signal.

it shifts the PCM to make it less loud, so if DN=1 the PCM you are getting out is not the original (correct) but one that is 30dB lower. On the other hand there is a reason, most times, that they will use a louder (or less loud) setting (i.e. something that will give a –1 instead of -31) for the normal, and that is that PCM has finite data (gradients), and this way it can have the range needed for what is important. But once the metadata is applied and the PCM distorted and sent out, what is at the extremes will get truncated (i.e. in the above example the lower 30dB won’t play any more).

Now chances are that it won’t be –1 and so not a 30dB change, and even at the noise floor in most peoples HT will be more then 30dB (let alone a more reasonable DN number then –1). But it definitely alters the PCM track and there definitely is an effect to what you hear

[ClaytonMG]

Originally Posted by Anthony P

But it definitely alters the PCM track and there definitely is an effect to what you hear

But it doesn't effect what you hear if you level match.

[Anthony P]

Originally Posted by ClaytonMG

But it doesn't effect what you hear if you level match.

yes it does

[BIslander]

Originally Posted by Anthony P

DN is important to DRC, I pointed out many pages ago that it was one of its use, but you don't really need it, in the end you need a 0 point (the point that does not move) where everything contracts or expands around. DN dictates that point based on the movie, but you could just as easily use any other point and DC around it.

Sure. But, if it's not DN, then what? Are you suggesting some other bit of metadata that defines the average dialog level? Isn't that already being done by the dialnorm value?

[ClaytonMG]

Originally Posted by Anthony P

yes it does

I the overal level of each of the chanels is decreased by 4db, and then increased by the end user to compensate, it doesn't matter.

[BIslander]

Originally Posted by ClaytonMG

DN is the only one that cannot be defeated and again IT DOES NOT ALTER THE PCM TRACK that is created during decoding.

Originally Posted by Anthony P

but no matter how many times you state that, it is still wrong. It does alter the PCM, if it did not then why would it be there in the first place.

Sorry, but repeating that "it's wrong" doesn't change the fact that the only change is the overall volume level. Dolby pegs the proper dialog level at -31dbFS while most movies are encoded at -27dbFS. Meanwhile, people watch movies at all sorts of volume levels. I hope you're not saying that someone isn't listening to lossless audio or watching a movie the way the director intended if they don't have their receiver set at the reference level used in the mixing room. Similarly, is someone "altering" the sound if he lowers the output of the player by 4db and then raises it by 4db in the receiver?

[Anthony P]

Originally Posted by BIslander

Sure. But, if it's not DN, then what? Are you suggesting some other bit of metadata that defines the average dialog level? Isn't that already being done by the dialnorm value?

no, Dolby's DRC uses the DN and that is what it is there for to create the "normal" (0 point) for that movie. What I am saying is that in general you don't need a normal based on the movie, you can have a fixed normal and use it for all movies. like we discussed before, on many devices even something that is not Dolby can and does not have DN or DRC can be DRCed, the only thing is that it will use a fixed point

[Anthony P]

Originally Posted by BIslander

Sorry, but repeating that "it's wrong" doesn't change the fact that the only change is the overall volume level. Dolby pegs the proper dialog level at -31dbFS while most movies are encoded at -27dbFS. Meanwhile, people watch movies at all sorts of volume levels. I hope you're not saying that someone isn't listening to lossless audio or watching a movie the way the director intended if they don't have their receiver set at the reference level used in the mixing room. Similarly, is someone "altering" the sound if he lowers the output of the player by 4db and then raises it by 4db in the receiver?

changing the volume does not matter, in the end there is no correct one, the same volume in a closet where your ears are an inch from the speakers won't appear as the same loudness as in an auditorium with the speakers 30' away.

It does not lower by 4dB (or what ever the DN shows it should be) the volume, it lowers by 4dB (or what ever the DN shows it should be) in the PCM, and then the receiver plays that back and you can raise it by 4 dB

maybe this simple example will help

1) Pick a number between 0 and 10 (original PCM)
2) now let's pretend we normalize it and so you subtract 4 (like your example) (normalization)
3) now that number still needs to be between 0 and 10 so if less then 0 then keep at 0 (PCM)
4) now we volume adjust to remove the effect of the DN so we add 4 (unless it is 0 (play back)

if your number was 8 then you get back 8, but if it was 0-4 step 3 (PCM) they all become the same value (0) and so they play back as the same value (0), no matter how much louder we add in 4 it will still be 0 because in step 3 we killed anything that was originally between 0 and 4.


The issue is that you and many others forget about step 3 the PCM and that anything below the threshold (difference between DN value and 31) will be killed.

[BIslander]

Originally Posted by Anthony P

no, Dolby's DRC uses the DN and that is what it is there for to create the "normal" (0 point) for that movie. What I am saying is that in general you don't need a normal based on the movie, you can have a fixed normal and use it for all movies. like we discussed before, on many devices even something that is not Dolby can and does not have DN or DRC can be DRCed, the only thing is that it will use a fixed point

So, you're suggesting guesswork? Pick a point and hope that it's the right 0 point for all sources. If it's not, then DRC doesn't work very well - you may end up compressing the dialog along with the effects or boosting it so that it's too loud, undermining the whole point of DRC in the first place. If there were some consistency with volume levels in the first place, there wouldn't be a need for dialnorm. It seems like DN is what makes DRC actually work in the real world. I must admit I'm confused about where you are going with this.

[BIslander]

Originally Posted by Anthony P

It does not lower by 4dB the volume, it lowers by 4dB in the PCM, and then the receiver plays that back and you can raise it by 4 dB

I'm not sure what that means. Are you saying that raising the master volume by 4db does not offset lowering the PCM by 4db?

maybe this simple example will help

Not really.

1) Pick a number between 0 and 10 (original PCM)
2) now let's pretend we normalize it and so you subtract 4 (like your example) (normalization)
3) now that number still needs to be between 0 and 10 so if less then 0 then keep at 0 (PCM)
4) now we volume adjust to remove the effect of the DN so we add 4 (unless it is 0 (play back)

if your number was 8 then you get back 8, but if it was 0-4 step 3 (PCM) they all become the same value (0) and so they play back as the same value (0), no matter how much louder we add in 4 it will still be 0 because in step 3 we killed anything that was originally between 0 and 4.

The issue is that you and many others forget about step 3 the PCM and that anything below the threshold (difference between DN value and 31) will be killed.

Let's add some reality to your simple example. How about using a scale of 0-105 and add in the fact that no one can hear anything from 0-4. So, what gets lost when we kill 0-4?

[Anthony P]

Originally Posted by BIslander

So, you're suggesting guesswork? Pick a point and hope that it's the right 0 point for all sources.

in a way yes, there is not that much variety between sources

If it's not, then DRC doesn't work very well - you may end up compressing the dialog along with the effects or boosting it so that it's too loud, undermining the whole point of DRC in the first place.

not at all,
1) let me ask you this Dolby considers –31 to be the normal

In contrast, a louder program requires a shift to match the –31 dB standard.

and things louder then –31 to be normalized, if something should give –51, and it DRCs with –31 because DN can’t be more then –31, why will it work and you don’t have an issue with it?

2) DRCing "too loud" does not make any sense, think about it, even if I compress everything and bring it "too loud", loudness depends of the volume setting on your device, if you crank it down then everything will be down low. And not move much and if you have it high everything will be high and not move much.

If there were some consistency with volume levels in the first place, there wouldn't be a need for dialnorm.

there is no consistancy because the scale (PCM) is fixed and so audio is digitized at the level to best match the "range" of the content

It seems like DN is what makes DRC actually work in the real world. I must admit I'm confused about where you are going with this.

for Dolby Yes, DN gives the reference point but like others pointed out and you discussed several pages ago many receivers do their own version of DRC and on sources that do not have DN. Will the sound be slightly different with out DN? possibly

[Anthony P]

Originally Posted by BIslander

I'm not sure what that means. Are you saying that raising the master volume by 4db does not offset lowering the PCM by 4db?

not uniformly

Let's add some reality to your simple example. How about using a scale of 0-105 and add in the fact that no one can hear anything from 0-4. So, what gets lost when we kill 0-4?

1) 0 is defined as the limit of average hearing not 4
2) in the end it depends on the volume knob, in the example we tried to remove the effect of DN, but someone in their home could want things louder then what Dolby considers reference (after all DTS considers it should be 4dB higher)
3) DN values range from -31 to -1, that means that it is not limited from 0 to 4 but could be 0 to 30
4) who are you to decide it does not and should not matter if there is a difference?
5) if you bothered to pay attention to my original post you would see that I said

Now chances are that it won’t be –1 and so not a 30dB change, and even at that the noise floor in most peoples HT will be more then 30dB (let alone a more reasonable DN number then –1). But it definitely alters the PCM track and there definitely is an effect to what you hear

so I agree that it is not a big issue, but the problem is it is there for anyone that wants to have an issue with it, and to pretend it does not exist is just being dishonest.

[BIslander]

So, with DRC you are saying guesswork is fine because sources don't vary much. Of course, that's not true. They're all over the map. This is about more than motion picture encodes on BD, although even they vary to some degree. It's about a wide variety of consumer entertainment audio.

Originally Posted by Anthony P

1) let me ask you this Dolby considers –31 to be the normal and things louder then –31 to be normalized, if something should give –51, and it DRCs with –31 because DN can’t be more then –31, why will it work and you don’t have an issue with it?

You think there's much recorded with the average dialog lower than -31dbFS? Care to offer some examples? But, yes, I agree it would be better if DN allowed for upward adjustment as well.

2) DRCing "too loud" does not make any sense, think about it, even if I compress everything and bring it "too loud", loudness depends of the volume setting on your device, if you crank it down then everything will be down low. And not move much and if you have it high everything will be high and not move much.

"Too loud" makes perfect sense. DRC lowers the volume above the 0 point and raises it below the 0 point. Average dialog plays at the recorded volume. Big bangs are reduced in volume so that the baby isn't awakened. And, whispers are increased in volume so that they can still be heard when watching at low volume in a quiet house. If the 0 point is too low, then dialog gets lowered along with the bangs. If it's too high, then dialog gets boosted along with the whispers. DRC only works correctly when the 0 point is correct.

for Dolby Yes, DN gives the reference point but like others pointed out and you discussed several pages ago many receivers do their own version of DRC and on sources that do not have DN. Will the sound be slightly different with out DN? possibly

I'd say definitely, which is consistent with the earlier posts you referenced.

But, again, where are you going with this? Using metadata to establish an accurate 0 point has no downside. Why do you prefer using guesswork that can never produce a better result and will often produce an inferior one?

[BIslander]

Originally Posted by Anthony P

1) 0 is defined as the limit of average hearing not 4

OK. And a whisper is at 15db. So, what are we actually losing between -101dbFS and -105dbFS?

2) in the end it depends on the volume knob, in the example we tried to remove the effect of DN, but someone in their home could want things louder then what Dolby considers reference (after all DTS considers it should be 4dB higher)

If you calibrate using a Dolby set of tones, which is the case with most calibration tools these days, reference will be the Dolby standard and DTS will be 4db too hot most of the time. But, no one listens at reference anyway. So, that's all quite irrelevant.

3) DN values range from -31 to -1, that means that it is not limited from 0 to 4 but could be 0 to 30

Sure. But, when average dialog gets that high, who cares. Losing the bottom end of an awful mix is no great loss, if that even happens.

4) who are you to decide it does not and should not matter if there is a difference?

???

5) if you bothered to pay attention to my original post you would see that I said "Now chances are that it won’t be –1 and so not a 30dB change, and even at that the noise floor in most peoples HT will be more then 30dB (let alone a more reasonable DN number then –1). But it definitely alters the PCM track and there definitely is an effect to what you hear"

I don't understand your point there either. PCM is altered in that the volume is lowered. And the impact of that? The potential loss of sounds that are too quiet to hear in the first place.

so I agree that it is not a big issue, but the problem is it is there for anyone that wants to have an issue with it, and to pretend it does not exist is just being dishonest.

Of course, it's there. No one is denying the existence of dialnorm. The discussion is about its effect. You acknowledge it's not a big issue. I'd say it doesn't matter for lossless playback and has value in other areas.

[Monkey]

Originally Posted by PeterTHX

Uh huh, that's some optimistic math there.

I figured you'd appreciate rounding up to 15% for Dolby Digital

[trans22]

Originally Posted by Monkey

I figured you'd appreciate rounding up to 15% for Dolby Digital

lol

[Anthony P]

So, with DRC you are saying guesswork is fine because sources don't vary much. Of course, that's not true. They're all over the map. This is about more than motion picture encodes on BD, although even they vary to some degree. It's about a wide variety of consumer entertainment audio.

come on, they don’t very that much, and it is not guess work, the same way that Dolby guessed –31 for reference average is not guesswork. The math also shows that it does not matter much

You think there's much recorded with the average dialog lower than -31dbFS? Care to offer some examples? But, yes, I agree it would be better if DN allowed for upward adjustment as well.

I don’t know, can you tell me what the average (normal) is for every movie you have seen? My guess you can’t even tell me one of them. I was just pointing out the obvious, that anything below what Dolby considers a reference norm will not have the right 0 for DRC, further more, this is the Sony thread and Sony has said they use DN=0, which means that DRC will be done as if it was DN=-31 even if it should be –21 or –41.

Too loud" makes perfect sense. DRC lowers the volume above the 0 point and raises it below the 0 point. Average dialog plays at the recorded volume. Big bangs are reduced in volume so that the baby isn't awakened. And, whispers are increased in volume so that they can still be heard when watching at low volume in a quiet house.

no it does not, it is volume only after it comes out of your receiver

If the 0 point is too low, then dialog gets lowered along with the bangs. If it's too high, then dialog gets boosted along with the whispers.

Yes but like with DN most of the fanciest receivers have this new fangled thing called volume control, and you can turn up or down the playback volume. You keep on saying DN does not change anything , if it is at –21 then DN will turn the “volume” down 10dB and if you want to pretend it does not exist then turn up your receiver by 10 dB. The same here, if you use DRC with that DN and it has 2:1 compression, then putting it down 5 dB will bring the dialogue to the same level as having applied DN before compressing. Furthermore the range will still be .5 of what it was originally.
[/quote] DRC only works correctly when the 0 point is correct.
[/quote]
not at all, and as I pointed out even with DN on Dolby the point does not need to be correct

But, again, where are you going with this? Using metadata to establish an accurate 0 point has no downside. Why do you prefer using guesswork that can never produce a better result and will often produce an inferior one?

no where, you are making a federal case of this, I was just pointing out that you don’t really need the real centre point in order to compress or expand the range of sound, just that Dolby decided to do it that way. And that is why range control can happen with no DN point.

[BIslander]

Sorry, Anthony, you've completely lost me here. Your responses also suggest you may not understand what I am saying. Perhaps someone else can translate.