Hi all, new here, but looks like a cool place to discuss audio.

Anyhoo, recently I've been working on post audio and mixing for a couple new television shows on FuelTV. We were told by the network that peak levels should stay below -8db, although I'm used to staying even lower with advertising, like -10db. So the masters (Dbeta) were created and we signed off on them in good spirits.

Well when watching the shows via broadcast, there were very noticeable changes in the overall level of the mix, like someone was riding a fader during the show!

After hearing this is a couple of episodes, I decided to record the audio stream from the broadcast and compare it to the original mix. My ears weren't lying after all, as there are very noticeable differences when visually comparing the waveforms. However, it definitely doesn't seem to be limiting, as you would see that clearly.

You can check out what I'm talking about with these comparison images. (They are just the L channel for quick comparison sake).





Anyone have any ideas of what may be going on here?

There's probably one (or more) automatic multi-band dynamics controllers in the audio paths somewhere -- Optimod-style devices. They aren't ise in the UK TV broadcast chain, but they are quite common in the USA distribution chains.

hugh

--------------------
Technical Editor, Sound On Sound

Quote Hugh Robjohns:

---

There's probably one (or more) automatic multi-band dynamics controllers in the audio paths somewhere -- Optimod-style devices. They aren't ise in the UK TV broadcast chain, but they are quite common in the USA distribution chains.

hugh

---

Right.. ok. I kinda figured it was something to that note. However, is there any way to change what I'm doing to get more predictable broadcast results? I mean, it's quite destructive.

They work by sensing the energy in different audio bands and try to level it out (which you can see they do pretty well). So the solution is to reduce the dynamics in your mix and try to keep the energy as even as you can.

Makes for a boring programme, though.

Personally, I prefer to mix to make it sound good on a proper linear system. If people complain at the end of a broadcast chain, then the complints can be directed at the poorly set up transmission chain -- not the source programme.

Frustrating, isn't it? As I said, we don't (yet) suffer this problem in TV in the UK, but it's been the bug-bear of radio for years.

hugh

--------------------
Technical Editor, Sound On Sound

Quote soundshaper:

---

...although I'm used to staying even lower with advertising, like -10db...

---

That's very interesting to hear. On this side of the pond I'm certain the opposite is true, with the adverts (commercials) appearing to be very much louder than the main program. Dunno about the peak levels but they are deliberately volume-maximized/compressed to make their RMS level higher. It's so obvious even your average Joe Bloggs writes to TV watchdogs complaining about it, but nothing ever gets done. It's another take on the continuing "loudness war".

--------------------
I'm a bedroom beatsmith in an 8x7 cell, writing trance dance and techno, and sometimes words as well...

Quote MajorFubar69:

---

On this side of the pond I'm certain the opposite is true, with the adverts (commercials) appearing to be very much louder than the main program.

---

The perceived loudness is often (still) very high, but the peak level (as seen on a waveform display) is actually significantly lower. Certainly for TV ads in the UK the ASA regulations (ASA rule 6.9) restricts the peak level of ads to 6dB below the peak level for programmes (ie, PPM4.5 instead of PPM6). It's a crude attempt to balance programme and ad levels, but it's never going to work very well.

Quote:

---

Dunno about the peak levels but they are deliberately volume-maximized/compressed to make their RMS level higher.

---

Yes, they are -- another example of the daft notion that louder means more people will hear the ad. In fact, a growing majority of people now mute their TVs to avoid the noise, so all that heavy compression has become largely self-defeating!

This is an inevitable problem when audio is metered by devices that look at the signal voltage. Thankfully, there is significant progress being made in introducing metering systems that do provide a reliable measurement of perceived volume, such as the ITU recommendations BS 1770 and 1771.

You might find this interesting:

www.ibs.org.uk/public/lineuparchive/2008/118_Nov-Dec/10_Hush_the_Ads.pdf

and this:

www.ibs.org.uk/public/lineuparchive/2008/118_Nov-Dec/11_DK-Technologies_Lo udness_Metering.pdf

Hugh

--------------------
Technical Editor, Sound On Sound

Thank you guys for discussing this with me.. very interesting stuff.

I guess a multi-band compressor or similar could definitely be the culprit, although in some instances it doesn't make sense. There's one bumper in the second example above that should be 2-3 dB louder than the following content, but instead it actually aired 3-4 dB lower than the following content. F'in crazy.

The rule of thumb I gathered from working at commercial recording studios is that advertising should not have peak levels of more than -10, so yes, sometimes engineers will squeeze the mix up into that peak pretty hard trying to get the maximum perceived loudness. Crossing above -10 dB is risking rejection by the networks. But for program(me) material, we usually mix with more dynamic range and a bit higher peak of -8dB.

But as I said, it really sounds like someone grabbed a master fader and pulled it down or even up at the strangest times. We were watching epidsode1 live at the premiere party and it sounded all over the place! Argghh! This network is also young and inexperienced, so maybe they have some multi-dynamics broadcast device set on a stupid preset or something. Very aggravating.

So is the answer to use multi-band compression myself, getting RMS equal among several bands? How would I know which frequencies are within each band, or the slope of the filters?

You're mastering too high.
-18 dBFS = 4 PPM
So your peak level should be -10 dBFS (=6PPM)
Anything above that will be limited.
You should aim for around -14 dBFS (5 PPM) and certainly never louder than -10.
This is industry standard.
When I'm mixing I don't hit 6PPM (-10dBFS) as I don't rely on limiters to do my job for me.
Going above that is very bad practice - you can't complain about the results, as I'm afraid you're to blame.
Never go above -10 dBFS.

There's nothing wierd happening - you're just going 2 dB higher than the industry peak level (-10 dBfs) and banging into their limiters.

Quote TVSound:

---

You're mastering too high.
-18 dBFS = 4 PPM
So your peak level should be -10 dBFS (=6PPM)
Anything above that will be limited.
You should aim for around -14 dBFS (5 PPM) and certainly never louder than -10.
This is industry standard.
When I'm mixing I don't hit 6PPM (-10dBFS) as I don't rely on limiters to do my job for me.
Going above that is very bad practice - you can't complain about the results, as I'm afraid you're to blame.
Never go above -10 dBFS.

There's nothing wierd happening - you're just going 2 dB higher than the industry peak level (-10 dBfs) and banging into their limiters.

---

Ok, that makes sense. Thanks.

They must have some awful settings on their equipment, cause what it's doing is drastic and extremely apparent.

Quote TVSound:

---

So your peak level should be -10 dBFS (=6PPM)
Anything above that will be limited.

---

Maybe... but unlikely, at least in the UK.

Quote:

---

You should aim for around -14 dBFS (5 PPM) and certainly never louder than -10.

---

There is a lot of confusion about acceptable peak audio levels in UK broadcasting, and part of that is the widespread lack of understanding of the differences between a standard PPM meter, and a standard digital meter.

The UK practice is to allow peaks up to PPM 6 as observed on a Type II PPM. A lot of people then simplistically extrapolate from the EBU line up of PPM4 = -18dBFS, and assume that PPM6 must equal -10dBFS. THIS IS NOT TRUE FOR DYNAMICALLY CHANGING PROGRAMME MATERIAL

The EBU alignment is performed with a steady state tone, but the acceptable peak programme level is metered with dynamically changing signal levels. This is a critical difference because the Type II PPM has a 10ms integration time -- it is not a true peak reading meter. In contrast, the vast majority of digital meters are true peak reading (they certainly have a zero integration time), and as a result will show brief signal peaks which the PPM will completely ignore (by design).

What this means in practice is that transient peaks that aren't seen on a standard PPM will be revealed clearly on a standard digital meter. As a result, a programme produced using traditional PPMs to peak no higher than PPM6 will, if checked using a digital meter, exhibit transient peaks up to -6dBFS or higher quite frequently. This is quite normal and totally expected -- and from a technical point of view, is actually completely acceptable too. Sadly, some broadcast companies currently have policy documents written by imbeciles that persist in demanding programmes do not exceed -10dBFS.

Quote:

---

There's nothing wierd happening - you're just going 2 dB higher than the industry peak level (-10 dBfs) and banging into their limiters.

---

Firstly, the OP is talking about his experiences in the US, which has a variety of different broadcast production standards, none of which are the same as the UK.

Secondly, the distribution chain in the US is considerably more complicated than that employed in the UK, and frequently involves the use of multiband compressors and mult-stage format conversions.

Third, the UK industry peak level IS NOT -10dBFS, although a lot of people (including some at very senior levels) share that particular gross misunderstanding -- you are certainly not alone.

And fourth, the kind of level control seen in those waveforms could not be caused by brick-wall transmission limiters.

Hugh

--------------------
Technical Editor, Sound On Sound

Quote soundshaper:

---

Ok, that makes sense.

---

Er... no it doesn't

hugh

--------------------
Technical Editor, Sound On Sound

Quote Hugh Robjohns:

---

And fourth, the kind of level control seen in those waveforms could not be caused by brick-wall transmission limiters.

---

This is why I have inquired with the network about their signal processing chain.

Although young, I've actually been mixing material for other networks for some time now, and I have never run into this issue before. I'll keep everyone posted on what we conclude, but for now I'll be observing a -10dB limit.

What exactly are broadcasters after in specifying and limiting peak levels, energy levels in various bands over certain integration times, etc.? A sense of perceived loudness that doesn't vary all that much? But why would broadcasters want that? Surely, you end up with strange things, like the rustling of clothing that's just as loud as footsteps, no?

By the way, what about films distributed on DVD? Do they typically have audio that uses the full dynamic range of the medium?

And what about projected films in theaters? Are they allowed to have audio that uses the full the dynamic range of the medium?

Thanks.

--------------------
Robert Eidschun

Hi Robert... to asnwer your questions in reverse order:

Quote Robert Eidschun:

---

what about projected films in theaters? Are they allowed to have audio that uses the full the dynamic range of the medium?

---

Yes. And many do. The thing that makes it possible is that cinemas have carefully specified, powerful and properly calibrated sound systems which are capable of delivering high sound levels when required, and reasonably well isolated theatres that allow the quiet bits still to be heard (assuming the popcorn and sweetie rustling is minimal )

Quote:

---

By the way, what about films distributed on DVD? Do they typically have audio that uses the full dynamic range of the medium?

---

DVDs generally do, yes, but becaues few home theatre systems could cope, the Dolby Digital sound track incorporates metadata that allows the decoder to reduce the dynamic range in a way which the film dubbing studio thought was acceptable -- if the home user activates it.

Quote:

---

What exactly are broadcasters after in specifying and limiting peak levels, energy levels in various bands over certain integration times, etc.?

---

Initially, it was to protect analogue transmitters from over-modulation and causing interference in adjaceant channels. It is also about providing consistent levels between programmes and channels so that the home viewer doesn't have to constantly adjust the listening volume.

In Europe, peak levels are set using a PPM meter with a nominal peak of +8dBu (UK) or +9dBu (most of Europe) -- but the important thing to realise is that the PPM isn't a true peak reading meter; it has a deliberately slugged response so that it ignores low-energy high-level transient peaks.

Quote:

---

A sense of perceived loudness that doesn't vary all that much?

---

Yes, but over the long term. It's not about making the sheep farts as loud as the airplane crashes -- that would clearly be stupid. It's about trying to maintain a consistent level for average conversational speech, so that if you change channels from watching a news report to a documentary to a soap opera to a children's programme, people talking normally all sound pretty much the same level as each other. And the same at programme junctions on the same channel. When one programme ends, the speech in the voice over announcement or adverts should be roughly the same volume as each other.

Quote:

---

Surely, you end up with strange things, like the rustling of clothing that's just as loud as footsteps, no?

---

Not at all.... but it is all about interpretation and common sense, and sometiems both appear to go AWOL!

hugh

--------------------
Technical Editor, Sound On Sound

Hi Hugh,

Thanks for your response. My comments are below, in-line.

Robert

**********

Quote Hugh Robjohns:

---

... cinemas have carefully specified, powerful and properly calibrated sound systems which are capable of delivering high sound levels when required ...

---

I'm not sure how particular such sound systems have to be other than loud enough. I was at Kilbourn Hall last week, where I heard the performance of a piece for Pro Tools and live voice: there were six Mackie HR824s, each fed a separate channel and arranged throughout the seating, and the sound was incredible and much better than I've ever heard in any cinema, with extremely quiet passages as well as thunderously loud sections. The composer sat in the middle of the seating, adjusting faders a wee bit now and then, but that's all he did.

Kibourn Hall, Eastman School of Music, Rochester, New York:

Quote Hugh Robjohns:

---

... the Dolby Digital sound track [on DVDs] incorporates metadata that allows the decoder to reduce the dynamic range in a way which the film dubbing studio thought was acceptable -- if the home user activates it ...

---

On my Sony Bravia TV, which has two built-in speakers, I do not listen to the Dolby track. In addition, that TV offers what is explicitly called "compression", but I keep that turned off. Consequently, I'm having to boost the quiet sounds (e.g., the rustling of clothing) of my own audiovisual project quite considerably so that they are audible, when my project is played back from DVD and shown (and heard) on that TV. But that means that if I listen to the soundtrack on my home stereo, the quiet sounds sound somewhat too loud. This is of concern because I can imagine that at certain film and video festivals, there may be hi-fi sound systems...

Quote Robert:

---

What exactly are broadcasters after in specifying and limiting peak levels, energy levels in various bands over certain integration times, etc.?

---

Quote Hugh Robjohns:

---

Initially, it was to protect analogue transmitters from over-modulation and causing interference in adjaceant channels.

---

Well, I assume that that is no longer an issue. It's hard to imagine that it would be in this day and age, given modern electronics.

Quote Hugh Robjohns:

---

It is also about providing consistent levels between programmes and channels so that the home viewer doesn't have to constantly adjust the listening volume.

---

I'm not sure I follow here... Do you mean, "so that the home viewer no longer 'wants' to adjust the volume"? If there is some standard specified by the broadcaster that correlates sound pressure level to signal level, then there shouldn't be a problem. So, for example, something like, "normal speach by a source within a few feet of the camera should produce average levels of around -x dBFS", or something like that.

Quote Hugh Robjohns:

---

In Europe, peak levels are set using a PPM meter with a nominal peak of +8dBu (UK) or +9dBu (most of Europe)

---

"+8 dBu" means that the voltage of the signal, v, is 8 db above 1 microvolt, i.e. +8 = 20 log (v/0.000001). But this specification is only meaningful in the analogue domain, where it refers to, for example, the voltage measured at the output of a playback head that is reading an analogue audio track off of tape. How then can such a specification pertain to an audio signal that is represented digitally?

Quote Hugh Robjohns:

---

-- but the important thing to realise is that the PPM isn't a true peak reading meter; it has a deliberately slugged response so that it ignores low-energy high-level transient peaks.

---

That probably means that it's showing the average signal level over time, using the Mean Value Theorem. So, one might ask, what is the area under a curve, over a certain width -- say, a curve that represents signal amplitude, expressed as voltage, over time? Answer: the integral over that width. But one might also ask, what is the height of a rectangle that has the same width and the same area? Answer: the integral just mentioned, divided by its width. That height is the "average" voltage, according to the Mean Value Theorem. The latter does nothing but validate the procedure that I just mentioned. But that procedure is only one way of specifying the average of a varying signal level. For example, one might determine the average power of a varying signal, since changes in power (rather than signal amplitude) correspond more closely to changes in perceived volume.

In practice, the PPM meter is probably much "dumber", not doing any calculus at all, but simply allowing a capacitor to charge up, or something like that, where the voltage reached across the capacitor would be a function of the waveform and for how long it was monitored.

Quote Robert:

---

Surely, you end up with strange things, like the rustling of clothing that's just as loud as footsteps, no?

---

Quote Hugh Robjohns:

---

Not at all.... but it is all about interpretation and common sense, and sometiems both appear to go AWOL!

---

I don't see how you won't end up with, as I pointed out, the rustling of clothing that's just as loud as footsteps, and in fact, that's what I've ended up with. The problem is that the scene in question is a classical music concert, and I've set things up so that the loudest part of the music is at full scale (and would thus clip if any louder). Now, when the performers whisper, it's barely audible, as it should be. But clothes rustling should be even quieter, and making them so results in their being inaudible. So, I've performed "manual compression", raising the volume of the rustling clothes. This seems OK until the music goes quiet, at which point you realize that the rustling clothes that you just heard were just as loud as the quiet music that you're now hearing, which seems a bit ridiculous. So, I raised the volume of the whispering too -- as much as I could stand before it started to seem too loud with respect to the rest of the music. This then allowed me to raise the volume of the rustling clothes, but which still ended up being too quiet. So, I raised the volume of the rustling clothes even more so, but now I'm back to having rustling clothes that are too loud with respect to the music.

I'm describing what I've done in order to accommodate the audio playback system of a standard TV. By contrast, if I were to set levels for audio playback on a good stereo, then I wouldn't need any "manual compression". There seems to be no good solution in the situation that I've described here, unless you have any ideas(?).

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

I'm not sure how particular such sound systems have to be other than loud enough.

---

Cinema sound systems are certified for levels, frequency response, dispersion, isolation and so on. THX certified cinemas are particularly tightly controlled. You can find the specs on the Dolby and THX websites.

Quote:

---

But that means that if I listen to the soundtrack on my home stereo, the quiet sounds sound somewhat too loud.

---

A perfect example of why the dynamic range control requirements are different for different reproduction systems in different listening environments. There is no 'one-fix-suits-all' solution. The Dolby Digital metadata (Dialnorm, DRC, downmixing ) is a very clever attempt to provide near-perfect results for different listening environments and situations, but it requires the production team to use it appropriately, and the home use to employ it in the approporiate situations. Many, like you, still avoid it and then wonder why they can't hear the quiet bits, or why the loud bits wake the babies!

Quote:

---

Well, I assume that that is no longer an issue. It's hard to imagine that it would be in this day and age, given modern electronics.

---

Peak level control will always be required, whether the broadcast chain is analogue or digital (or hybrid). As long as analogue broadcasting continues (as it will here for another few years in TV, and indefinitely in radio), there will remain the need to prevent overmodulation, and digital systems must snever be allowed to clip, obviously.

The advantage of digital systems is that the alignment of different stages and interfaces becomes far simpler, but peak lvels still need to be monitored and controlled.

Quote:

---

I'm not sure I follow here... Do you mean, "so that the home viewer no longer 'wants' to adjust the volume"?

---

No, I mean so they don't feel the need to adjust the volume between programmes or channels. In an ideal world, the home viewer/listener should be able to set the listening level they feel comfortable with and leave it at that for the duration of their viewing. They shouldn't have to turn it down when the adverts come on during the programme they are watching, or adjust it significantly when they switch to a different channel to watch a different programme.

...but this is a surprisingly difficult thing to achieve and is actually far more complicated than it appears. Some very clever peoploe in the EBU and other standards bodies have been trying to find ways of addressing it for years.

Quote:

---

If there is some standard specified by the broadcaster that correlates sound pressure level to signal level, then there shouldn't be a problem.

---

There isn't and can't be. There is in the film/cinema world, because the replay environment is known and the signal level/SPL level can be defined and fixed -- that's part of the Dolby/THX spec.

But in the home environment, there is no fixed listening level. Some people listen loudly, others quietly. Some have noisy backgrounds to content with (eg, the kitchen ) and some need to keep it quiet to avoid waking the children. So in general, the dynamnic range must be kept relatively small. The BBC's own guidelines suggest maintaining dialogue within an 8dB range, for example.

Quote:

---

So, for example, something like, "normal speach by a source within a few feet of the camera should produce average levels of around -x dBFS", or something like that.

---

That is done, as I said above. Normal speech would typically register between PPM4 and PPM6 on a BBC-style peak meter, which is roughly -18 to -10dBFS.

Quote:

---

"+8 dBu" means that the voltage of the signal, v, is 8 db above 1 microvolt, i.e. +8 = 20 log (v/0.000001).

---

No. 0dBu is a reference voltage of 0.775V (RMS). So +8dBu is a signal level of 1.95V (RMS). The 'u' in dBu is derived from 'unterminated' and refers to the fact that modern audio interfaces are no longer the matched impedance types that were used up until about 40 years ago. Before then, the standard audio signal level was 0dBm where 'm' referred to the power 1 milliwatt. The standard citcuit impedance was 600 ohms, and dissipating 1mW in 600 Ohms produces a voiltage of 0.775V. Modern interfaces are voltage-interfaces, and we kept the same reference voltage.

Many US audio systems use an elevated reference voltage of +4dBu, or 1.223V (RMS).

Quote:

---

But this specification is only meaningful in the analogue domain

---

Obviously, because it is an analogue voltage signal measurement.

Quote:

---

How then can such a specification pertain to an audio signal that is represented digitally?

---

In the digital domain we use dBFS (dB below full scale), and the relationship between dBu and dBFS is set by the A-D and D-A converters employed. The EBU recommends an alignment of 0dBu = -18dBFS, while the SMPTE recommends +4dBu = -20dBFS.

Quote:

---

Answer: the integral over that width.

---

Yes, pretty much all analogue meters are integrating meters. While the VU meter is intended to give some indication of 'avearage signal level' which has a vague relationship to perceived volume, the European PPMs were designed to give a much better and more reliable indication of signal peak levels.

However, a very deliberate engineering decision was made to prevent the meter from displaying fast transient peaks since even if these were allowed to overdrive transmitters, the resulting harmonic distortion products were too brief to cause annoyance, and leaving sufficient headroom to avoid transient overloads undermodulated the system very wastefully. To that end, European PPMs integrate over 5 or 10ms, and that's why they don't indicate fast transient peaks.

That was fine in the analogue domain, but is no longer entirely appropriate for the digital world where distortion caused by transient overloads in digital systems results in anharmonic distortion which is instantly audible, even when very brief. But the industry's familiarty with analogue PPMs ensures their continued use in many areas, and so the digital systems havea been engineered to maintain adequate headroom to prevent transient overload (as detailed earlier).

Quote:

---

one might determine the average power of a varying signal, since changes in power (rather than signal amplitude) correspond more closely to changes in perceived volume.

---

This is true, but reliable determination of perceived volume is quite complicated, and a lot of work has been (and is being) done on finding a consistent way of metering perceived volume. Dolby produce a bespoke system for metering perceived loudness (primarily for the film industry) and a lot of broadcasters are now adopting the ITU Recommendations BS.1770 and BS.1771 for loudness metering. Meter manufacturers such as DK-Technologies and RTW already manufacturer suitable devices, and I think this will become the way forward in maintaining consistent programme and channel signal levels.

Quote:

---

In practice, the PPM meter is probably much "dumber", not doing any calculus at all

---

I think you'd be surprised just how complicated the electronics of an analogue PPM is -- and there definitely is calculus in the signal processing.

Quote:

---

I don't see how you won't end up with, as I pointed out, the rustling of clothing that's just as loud as footsteps, and in fact, that's what I've ended up with.

---

Rustling clothes can certainly be as loud as footstpes, or footsteps as quiet as rustling clothes -- it all depends on the clothes, the shoes and the surfaces!

But I take your point. At the end of the day it comes down to the skill and artistry of the dubbing mixer to balance the sounds as appropriate withing the dynamic range constraints of the medium he or she is working in.

The key is to maintain a consistent dialogue level, that matches the guidelines of the format, and then balance everything else around that, within the dynamic range constraints of the medium.

Quote:

---

The problem is that the scene in question is a classical music concert, and I've set things up so that the loudest part of the music is at full scale (and would thus clip if any louder). Now, when the performers whisper, it's barely audible, as it should be.

---

I'll take your word for it. There is such a thing as a 'stage whisper' which gives the impression of whispering but remains entirely audible. There is little point in giving a performance where a proportion of the audience can't hear what is being said! And what might be barely audible in a cinema or oonn a hiigh quality dueatre monitoringg system will be completely inaudible when vieewed on a typical domestic TV. Hence the inherent need for skillful dynamic range control to suit the replay environment.

Quote:

---

So, I raised the volume of the rustling clothes even more so, but now I'm back to having rustling clothes that are too loud with respect to the music.

---

No one said the art of audio dubbing was an easy one! Mixing (or at least regualrly checking a mix) on a comparative monitor system is essential until you have acquired sufficient expereince to know how a mic will translate from big hifi monitors to small TV ones. And mixing at low listening levels will also help to produce more appropriate and consistent results.

Quote:

---

There seems to be no good solution in the situation that I've described here, unless you have any ideas(?).

---

It's a compromise, quite obviously. Some aspects of mixing perfection have to be compromised to produce a mix which is satisfactory when heard in a typical domestic environement on a typical TV. That's just the way it is. Skillful balancing can make that compromise less obvious, but it will always be there.

hugh

--------------------
Technical Editor, Sound On Sound

Hugh,

"Thanks for your additional, very helpful comments. My comments are below, in-line.

Robert

**********

Quote Robert:

---

Well, I assume that that is no longer an issue. It's hard to imagine that it would be in this day and age, given modern electronics.

---

Quote Hugh:

---

Peak level control will always be required, whether the broadcast chain is analogue or digital (or hybrid). As long as analogue broadcasting continues (as it will here for another few years in TV, and indefinitely in radio), there will remain the need to prevent overmodulation, and digital systems must snever be allowed to clip, obviously.

---

I'm not sure what you mean by overmodulation, as the audio that accompanies an analogue TV signal is FM modulated, and one needs only keep frequency modulation of the audio carrier below a certain value. (I don't know what the maximum allowable modulation frequency is for analogue TV, but for FM radio in the USA, it's about 17 kHz.) However, modulation, as I just described it, is independent of amplitude, and yet your original comment was that one had to control amplitude to prevent overmodulation. Perhaps I'm missing something here.

Quote Robert:

---

I'm not sure I follow here... Do you mean, "so that the home viewer no longer 'wants' to adjust the volume"?

---

Quote Hugh:

---

No, I mean so they don't feel the need to adjust the volume between programmes or channels... They shouldn't have to... adjust it significantly when they switch to a different channel to watch a different programme.

---

Quote Robert:

---

If there is some standard specified by the broadcaster that correlates sound pressure level to signal level, then there shouldn't be a problem.

---

Quote Hugh:

---

There isn't and can't be.

---

Quote Robert:

---

So, for example, something like, "normal speach by a source within a few feet of the camera should produce average levels of around -x dBFS", or something like that.

---

Quote Hugh:

---

That is done, as I said above. Normal speech would typically register between PPM4 and PPM6 on a BBC-style peak meter, which is roughly -18 to -10dBFS.

---

Hugh, I intended my first statement ("correlates sound pressure level to signal level...") to be a generalization of the example given in my second statement ("normal speach...")

Quote Robert:

---

+8 dBu" means that the voltage of the signal, v, is 8 db above 1 microvolt, i.e. +8 = 20 log (v/0.000001).

---

Quote Hugh:

---

No. 0dBu is a reference voltage of 0.775V (RMS).

---

Actually, it's different here in the States, where we use "dBv" to mean your "dBu", and "dBu" here means yet something else. But my point here was only that such a figure alone does not specify the magnitude of a digital representation of a signal, despite all of the suggestions to the contrary earlier in this thread.

Quote Robert:

---

In practice, the PPM meter is probably much "dumber", not doing any calculus at all

---

Quote Hugh:

---

I think you'd be surprised just how complicated the electronics of an analogue PPM is -- and there definitely is calculus in the signal processing.

---

That means that it's sampling and then using a microprocessor to do maths. By the way, this isn't necessary, as you can always use an "analogue computer" to function as a meter, as was done in the past. So, for example, rather than calculating how much time is required to charge a capacitor, you simply charge a capacitor and measure how long it took.

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

I'm not sure what you mean by overmodulation

---

the origina of the problem relates to early Am broadcasting, but over-deviation of FM broadcasts is equally anti-social.

Quote:

---

Actually, it's different here in the States

---

Ah... two nations separated by a common language!

Quote:

---

But my point here was only that such a figure alone does not specify the magnitude of a digital representation of a signal, despite all of the suggestions to the contrary earlier in this thread.

---

In the strict sense, you are right, but the accepted EBU and SMPTE conventions enable people to assume a fixed relationship between analogue signal levels and their digital representation in a usefully consistent way.

Quote:

---

That means that it's sampling and then using a microprocessor to do maths.

---

No. It's quite possible to perform quite involved maths functions with analogue electronics. I suggest you google-search for the standard PPM circuits to see how involved it is.

Hugh

--------------------
Technical Editor, Sound On Sound

Quote Hugh:

---

No. It's quite possible to perform quite involved maths functions with analogue electronics. I suggest you google-search for the standard PPM circuits to see how involved it is.

---

Let me clarify what I was on about... Mathematics is a human construct that involves numbers, and because of that, there is never any maths done outside of humans unless numbers are explicitly represented -- never any maths done by analogue electronics, and in general, never any maths done outside of humans if humans have not developed numerical algorithms and implemented them in machines, where inside those machines, there are explicit representations of numbers.

This might seem merely philosophical and not terribly practical, but it is what distinguishes analogue from digital electronics: in the latter, there are explicit representations of numbers in the hardware, where as in the former, there is an analogue of the physical process being analyzed.

So, for example, we can determine the frequency at which a circuit oscillates by using a digital computer to solve numerically the differential equation that describes that circuit; or, we can use an analogue computer instead, switching in the various circuit elements to create physically the circuit in question, then apply a voltage to its input, then observe at which frequency it oscillates. Since we know which differential equation corresponds to that circuit, the frequency that we observe is the same as the frequency sought in solving the equation using mathematics, even though no mathematics is performed by the analogue circuit.

As an even simpler example, imagine that you want to determine the speed reached by a ball after it has fallen from rest in a vacuum for a certain amount of time. You could simply multiply g and t, where g is the acceleration due to gravity and t is the elapsed time; or, you could actually drop a ball in a vacuum, let it fall for a certain amount of time, then measure its speed. Now, clearly, in using the latter method, neither the ball nor you have multiplied g and t, or have done any maths whatsoever. You, your watch, the ball, etc. constitute the analogue computer; your brain, engaged to multiply g and t, constitutes the digital computer.

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

Let me clarify what I was on about...

---

OK. I take the point, even if it is philosphical pedantry that isn't relevant to the thread or the point I was making. But... er... thanks.

hugh

hugh

--------------------
Technical Editor, Sound On Sound

I didn't really mean to sound that way -- only to illustrate the difference between two distinct ways of using electronics to determine the same quantity: through the explicit representation and manipulation of numbers to solve mathematical equations versus the observation of systems whose behavior is analogous to the behavior of those equations.

It is easier to build the latter, but more difficult to design, since you have to know the maths to some degree of rigor, such as that associated with analytical integration, e.g., the Fundamental Theorem of calculus. By contrast, integration in the digital domain requires only geometry, and more specifically, the notion of determining the area under a curve by adding up the areas of a finite number of rectangles, etc.

Then again, the multiplicity of digital circuits in any given digital machine makes for very intricate design -- but it doesn't require the abstract notions of analytical mathematics, which most folks find much more difficult than Boolean algebra.

On another note, after much "manual compression" (through manual, upward adjustment in volume of the quiet bits), the dynamic range of my soundtrack is now 45 dB. Is that small enough for a low cost TV, with built-in speakers, etc.? By the way, the dynamic range seems adequately small on a couple of such TVs that I have, as the quietest bits are just barely audible, while the loudest bits are still below the level at which I think most folks would say, "the TV is too loud -- turn it down!"

(Of course, the loudest bits are at full scale, i.e. sixteen one's, given that it's 16 bit, 48 kHz audio, designed for DVD.)

Thanks.

--------------------
Robert Eidschun

Quote Hugh Robjohns:

---

Quote Robert Eidschun:

---

Let me clarify what I was on about...

---

philosphical pedantry

---

Hugh...where can I get one, what does it do and how much does it cost?

--------------------
My head hurts!

Quote Hugh:

---

Normal speech would typically register between PPM4 and PPM6 on a BBC-style peak meter, which is roughly -18 to -10dBFS.

---

Hi Hugh,

Just as a reminder, I'm working on the audio track for something to be distributed on DVD, primarily to be entered into film competitions, and I'd like the loudest bits to be at full scale.

I have a few more questions for you then, if you would be so kind as to provide answers please, so that I can be off and on my way:

1) In reference to your remark above, do you mean that the average peak value should be between the limits that you've specified, or that all peak values should be between those limits? If the latter, then this might require not only amplification or attenuation, but compression as well, in order to keep all of the speech entirely between those limits at all times. The procedure then would be first to compress the speech to a dynamic range of 8 dB, if necessary, then amplify or attenuate to move the speech into the range that you've specified. Have I got that right?

By the way, I have a habit of speaking with widely varying volume, even within a single sentence, and in listening to a recording of myself speaking normally, I notice it even more. It seems a bit excessive, even to the extent of making it a bit difficult to understand me, I imagine. So, I can certainly imagine using compression on a recording of my voice, for the sake of intelligibility.

2) The level of everything else other than normal speech should be set with respect to normal speech, i.e. with respect to the range that you've specified for normal speech. So, for example, thunder might be set at full scale, while the rustling of clothing would probably be set at something less than -18dBFS. Do I have that right?

3) After much "manual compression" (through manual, upward adjustment in volume of the quiet bits), the dynamic range of my soundtrack is now 45 dB. Is that small enough for a low cost TV, with a built-in speaker or two? Indeed, the dynamic range now seems adequately narrow on a couple of such TVs that I have, as the quietest bits are now just barely audible, while the loudest bits are still below the level at which I think most folks would say, "the TV is too loud -- turn it down!"

Thanks.

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

the dynamic range of my soundtrack is now 45 dB. Is that small enough for a low cost TV, with built-in speakers, etc.?

---

I don't think you'll find anywhere a definition of an acceptable dynamic range. In a more practical sense, the BBC PPM has as working range of about 24dB, and normal practice is to keep the peaks of anything critical within that metered range. The total system dynamic range for analogue TV is around 80dB. Film soundtracks tend to have a wider dynamic range because they are intended for viewing in an entirely different environment, but you can't have it both ways (unless you take advantage of Dolby Digital's metadata.

Quote:

---

the dynamic range seems adequately small on a couple of such TVs that I have, as the quietest bits are just barely audible, while the loudest bits are still below the level at which I think most folks would say, "the TV is too loud -- turn it down!"

---

At the end of the day, that's what matters. It is a combination of specific technical criterion, and aesthetic acceptibility. If you and your clients are happy with the mix, and it falls within the appropriate tech specs, then it's fine.

Quote:

---

(Of course, the loudest bits are at full scale given that it's 16 bit, 48 kHz audio, designed for DVD.)

---

There's no 'of course' about it. It comes down to standards again. Films generally do exercise the full dynamic range available and extend to full scale on occasions. TV programmes (both broadcast and on DVD), in general, do not, as in my experience they typically maintain the broadcasting standards of restricting peaks to around -9dBFS.

You have to decide: is this primarily a programme to be viewed in a cinema, in which case it should be mixed to cinema standards, or for TV, in which case it should be mixed to TV standards.

The only saving grace is that if you mix it to filmic stanards, you could then take advantage of the metadata in the Dolby Digital soundtrack that I presume will be used ont he DVD, to establish suitable dynamic range reduction characteristics to make the mix sound more acceptable (and match the regulations) on TVs.

Hugh

--------------------
Technical Editor, Sound On Sound

Quote Robert Eidschun:

---

Just as a reminder, I'm working on the audio track for something to be distributed on DVD, primarily to be entered into film competitions, and I'd like the loudest bits to be at full scale.

---

Fair enough. But don't be surprised when it sounds much louder than other TV programmes (unless you implement Dolby Digital metadata properly).

Quote:

---

If the latter, then this might require not only amplification or attenuation, but compression as well, in order to keep all of the speech entirely between those limits at all times.

---

Normal dubbing does involve all that yes. The best advice is to analyse commercial films (and/or TV programmes) on DVD that you think sound good, and see how they have handled the dialogue.

In the UK, TV dialogue is normally contained between about PPM 4 and PPM 6 -- roughly -18 to -10dBFS. In film mixes, the dialogue is generally lower (to allow more headroom for the explosions), typically around -24 to -12dBFS. But these are only guides, and it is the way the mix sounds that matters.

Quote:

---

The procedure then would be first to compress the speech to a dynamic range of 8 dB, if necessary, then amplify or attenuate to move the speech into the range that you've specified. Have I got that right?

---

That is certainly one way of dealing with it... but I'm not suggesting that the quietest whisper should only be 8dB quieter than the loudest scream. I'm talking about the general normal dialogue which acts as the benchmark for other level judgements. Stuff that's meant to be loud can be loud, and stuff that's meant to be quiet can be quiet. The skill and art is in achieving a mix that enables everything to be followed without being difficult to endure or for the mix to draw attention to itself.

Quote:

---

2) The level of everything else other than normal speech should be set with respect to normal speech, i.e. with respect to the range that you've specified for normal speech. So, for example, thunder might be set at full scale, while the rustling of clothing would probably be set at something less than -18dBFS. Do I have that right?

---

Yes. Context has a lot to do with it too, of course. In a dramatic, tense scene, for example, you might allow everything to get quieter through the duration of the scene in order to draw the audience in as well as to provide more 'subjective headroom' to accommodate the sudden high-impact shock of a huge explosion of terrifying scream at the climax of the scene. A mix is a dynamic thing that evolves and develops along with the film.

Quote:

---

...the dynamic range of my soundtrack is now 45 dB. Is that small enough for a low cost TV, with a built-in speaker or two?

---

It depends how you're defining dynamic range, but possibly. I don't know the material, but as I said vefore, the BBC PPM (as used for judging TV balances) has a dynamic range on its scale of about 24 dB and in most cases, a balance engineer wouldn't let the needle sit on the bottom end stop for too long unless there was a really good reason.

Quote:

---

Indeed, the dynamic range now seems adequately narrow on a couple of such TVs that I have, as the quietest bits are now just barely audible, while the loudest bits are still below the level at which I think most folks would say, "the TV is too loud -- turn it down!"

---

You're probably too close to the project to really make that call. I would suggest burning a disc and letting a few friends/family/colleagues have a listen and see how they respond to it. They'll tell you if it meets their expectations and acceptibility factors or not.

Hugh

--------------------
Technical Editor, Sound On Sound

Thank you very much for your comments, Hugh. Below are a few follow-up questions, as well as comments.

Robert

Quote Hugh:

---

Films generally do exercise the full dynamic range available and extend to full scale on occasions.

---

Just to make sure we're using the same terminology here, if you're using the full dynamic range of the medium, then your maximum level is at full scale, right? (And interpreting "full dynamic range" literally, you could say that the minimum level is at "minimum scale", e.g., 0000 0000 0000 0001 in 16-bit binary form.)

Quote Hugh:

---

The only saving grace is that if you mix it to filmic stanards, you could then take advantage of the metadata in the Dolby Digital soundtrack that I presume will be used ont he DVD...

---

I am not, in fact, going to include such a metadata track, as I've heard that the audio playback system at a film festival can be pretty bad, and so I'm afraid that the folks running the "show" wouldn't know how to make the appropriate audio selection, or wouldn't bother to make any such selection, but instead would just press "play". One composer I know, who writes for film, says that he makes two versions of his work, one compressed relative to the other, but has each put on a different DVD. Just before screening time and after he has had a chance to evaluate the audio playback system in use, he tells the "projectionist" which DVD to play.

Quote Hugh:

---

In film mixes, the dialogue is generally lower (to allow more headroom for the explosions), typically around -24 to -12dBFS.

---

I see. That, in particular, was something that I needed to know. Thanks.

Quote Robert:

---

...the dynamic range of my soundtrack is now 45 dB. Is that small enough for a low cost TV, with a built-in speaker or two?

---

Quote Hugh:

---

It depends how you're defining dynamic range...

---

Well, my track is normalized, and so the largest sample value is 2^16 (two to the power of sixteen) minus one (which is the maximum value of a 16-bit word, i.e. 1111 1111 1111 1111 in binary form) and the smallest sample value is 45 dB down from that:

-45 = 20 log (x/[(2^16) - 1]), where x is the smallest sample value.

So, x is about 369 (in base 10).

By the way, the background ambience (which is present during the shots when no music is playing) and the very light rustling of clothing are the quietest sounds, and I intend for those sounds to be barely audible. That is indeed the case if, during playback on a TV, you have the volume adjusted such that the loudest bits are just shy of being obnoxious. However, as I mentioned in a previous posting, if you play back the soundtrack on a good stereo and adjust the volume in the same manner (so that the loudest bits are just shy of being obnoxious), then the quietest bits seem a little too loud, but still plausible, I suppose.

My project incorporates opera and other kinds of classical music, along with "contemporary-classical" music, acted out in some cases, and upon which the drama is sometimes totally dependent. In fact, you could almost call the project one long music video. But I don't want to use the full dynamic range of 16-bit audio (~96 dB), even if that's done on certain DVDs and in films projected in cinemas, because I suspect that the audio playback systems at film festivals would not be able to handle that. Moreover, I believe that preliminary evaluation at such festivals, conducted simply to determine if an entry should be accepted or not, is done by judges watching DVDs on televisions at home, which is why my soundtrack has to have a dynamic range suitable for such a situation.

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

Just to make sure we're using the same terminology here, if you're using the full dynamic range of the medium, then your maximum level is at full scale, right? (And interpreting "full dynamic range" literally, you could say that the minimum level is at "minimum scale", e.g., 0000 0000 0000 0001 in 16-bit binary form.)

---

I very much doubt anything within the programme really gets down to -96dBFS in practuice, but in theory it could, and yes, full scale is full scale.

Quote:

---

I am not, in fact, going to include such a metadata track

---

If the audio on the DVD is encodeed in Dolby Digital (and most are) then the matadata track is there anyway and needs to have some sensible values applied. The dialnorm, downmix and DRC metadata doesn't apply to filmic presenations, but would be relevant to TV viewers.

I think you'd find a lot of useful information on the www.dolby.com website. There are several white papers covering all aspects of mixing for film.

hugh


Hugh

--------------------
Technical Editor, Sound On Sound

Quote:

---

If the audio on the DVD is encoded in Dolby Digital

---

I was planning to use no encoding at all -- just 16-bit, 48 kHz PCM audio.

Quote:

---

the matadata track... needs to have some sensible values applied.

---

I still can't conceive of what this is all about. You mean you specify values for compression during the DVD authoring process, and then during playback, the viewer can select from among those values? But when I insert a DVD into any of my five DVD players, I am presented only with the opportunity to enable "English Dolby 5.1 Surround Sound", and who knows what kind of mix results? A stereo mix? A mix comprising more than two channels? A compressed mix? A mix with "Dialogue Normalization", as Dolby calls it? After all, my DVD players cannot possibly detect what kind of TVs are connected, since those connections are just analogue component or composite video.

And if it is somehow the DVD player that creates the appropriate mix, then why does my Sony Bravia flat screen TV seem to indicate, by the words, "Dolby 5.1 Surround Sound capable", appearing a certain sticker, that it is involved in the decoding process?

Quote:

---

I think you'd find a lot of useful information on the www.dolby.com website. There are several white papers covering all aspects of mixing for film.

---

I hunted around that site for about 15 minutes and found mainly marketing information, no white papers, not a single mention of compression, etc. Can you indicate a particular page on that site that has sufficiently detailed technical information?

Based on what I see when I insert a DVD in my DVD players, I see no way to select different levels of compression in order to accomodate the playback equipment at hand. I can just imagine handing a DVD with a Dolby soundtrack to someone at a film festival who inserts it into a DVD player and then simply presses "play", upon which an uncompressed mix is played, despite the playback device being nothing more than an ordinary TV, in which case a great deal of the soundtrack might be inaudible. I've already heard that the audio playback systems at such festivals are poor, and that entries are played for audiences from DVD, on nothing more than 1080i wide screen TVs, etc. I can just see some non-technically inclined person operating the DVD player...

By the way, ever since 2002, which is when I started watching feature films on DVD at home, I always made sure that the Dolby soundtrack was de-selected in the set-up menu (in those cases where it was an option because there was an alternative soundtrack available too) because I didn't understand what it was and figured that you needed special playback equipment in order to decode and hear it properly; and in those cases, I never had the impression that the dynamic range was too great, despite watching on televisions with mediocre sound. I suppose that the track that I was listening to had most everything within a narrow range -- say, 45 dB, which is what I've found as the maximum that works for a crummy TV with one small speaker.

And funny enough, on Sunday evening, I watched a DVD that had only a Dolby soundtrack and its dynamic range was clearly too great for my Sony Bravia flat screen TV: either the loud bits were too loud, or the quiet bits were too quiet, thus forcing me to adjust constantly the volume during playback, which was very annoying. I cannot risk that happening during playback of my project.

I would surmise that 99% of the consumers out there do not bother making the appropriate soundtrack selection, Dolby or otherwise, when watching DVDs; that 95% of the directors and producers don't bother either, since most are not technically inclined; and that most of the film festival "projectionists" don't do so. It sounds like the whole Dolby thing is mainly for playback in commercial cinemas and for commercial broadcast, and mainly a marketing ploy for other contexts, where it sometimes even results in worse audio, as was the case for me on Sunday evening.

We are all extremists on this board, relatively speaking, indulging in rarefied practices in order to do our best, and thus willing to set up this and that for optimum sonic reproduction. Most of the people out there -- even those in the business -- will not exert themselves to the same extent.

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

literally, you could say that the minimum level is at "minimum scale", e.g., 0000 0000 0000 0001 in 16-bit binary form.

the largest sample value is 2^16 (two to the power of sixteen) minus one (which is the maximum value of a 16-bit word, i.e. 1111 1111 1111 1111 in binary form)

---

I always thought that most digital audio systems used two's complement binary. Must've got myself confused somewhere!

Very interesting thread. I learned a lot.
Thanks guys

--------------------
It wasn't me!
(Well, actually, it probably was)

Quote rhd_toyl:

---

I always thought that most digital audio systems used two's complement binary. Must've got myself confused somewhere!

---

"Two's complement" refers to two's complement multiplication, which is a means of accomplishing multiplication (in binary) through addition, through the shifting of bits before addition, etc. It was largely in use before microprocessors were able to multiply internally, and before the implementation of any assembly language command to do so. But since the mid-1980s, most microprocessors have included a multiply command -- although, for all I know, two's complement multiplication may still be carried out by the microprocessor as a way of executing such a command.

--------------------
Robert Eidschun

Thanks for explaining that - I thought two's complement binary was a number system where the msb represented the polarity of the following bits. I'm actually starting to regret weighing in on something I don't know very much about. . .

Quote Robert Eidschun:

---

I was planning to use no encoding at all -- just 16-bit, 48 kHz PCM audio.

---

Fair enough.

Quote:

---

You mean you specify values for compression during the DVD authoring process, and then during playback, the viewer can select from among those values?

---

No. The viewer configures their DVD or home theatre system to either ignore the values for the full cinema experience, of apply them for a more restrained performance. The metadata simply records a value that determines one of many pre-set dynamic range reduction curves that the dubbing mixer decided was the most appropriate for the material. Some systems will allow a user-variable degree of 'amount of squash' to be selected, but the overall dynamic control curve is preset by the dubbing mixer who defines the metadata.

It's a slightly involved system that's obviously not relevant to you in this case. But you'll find plenty of articles explaining it on the Dolby website. Search for Dolby Digital and metadata.

Quote:

---

But when I insert a DVD into any of my five DVD players, I am presented only with the opportunity to enable "English Dolby 5.1 Surround Sound"

---

Usually there is a setting to enable Dolby's DRC somewhere in a set up menu, or the analogue outputs are preset to always apply dynamic range reduction.

Quote:

---

, and who knows what kind of mix results? A stereo mix? A mix comprising more than two channels? A compressed mix? A mix with "Dialogue Normalization", as Dolby calls it?

---

Again, if you configure the machine (or home theatre system) to provide a stereo mix from a Dolby 5.1 soundtrack, the downmix parameters are defined by the dubbing mixer and stored in the metadata. The dubbing mixer can determine at what level (from a range of preset options) the centre and surround channels are mixed into the front left and right.

Dialnorm is always applied and is essentially a remotely preset volume control (determined yet again by the dubbing mixer) that allows the dialogue from films and TV shows and other material on DVD to sound roughly the same loudness.

Quote:

---

After all, my DVD players cannot possibly detect what kind of TVs are connected, since those connections are just analogue component or composite video.

---

Which is why most DVD's have set-up menus so that you can determine the picture format (4:3, 16:9, widescreen, letterbox, PAL, NTSC, progressive, RGN, Component, stereo, 5.1, DRC, screensaver and all the rest.....

Quote:

---

And if it is somehow the DVD player that creates the appropriate mix, then why does my Sony Bravia flat screen TV seem to indicate, by the words, "Dolby 5.1 Surround Sound capable", appearing a certain sticker, that it is involved in the decoding process?

---

The decoding of Dolby digital can be done in the DVD player, to present 5.1 or stereo analogue outputs (or via the SCART connections), or if the raw datastream is passed via S/PDIF (coax or optical) to a home theatre controller, that unit can handle the decoding instead.

And if the TV has a receiver for digital terrestrial or satellite, it may well have a Dolby Digital decoder itself.

Again, there's lots and lot's about this -- with explanations at all levels and for all audiences -- on Dolby's website.

Quote:

---

I hunted around that site for about 15 minutes and found mainly marketing information, no white papers, not a single mention of compression, etc. Can you indicate a particular page on that site that has sufficiently detailed technical information?

---

Come on Robert. It's a superbly designed site and very easy to navigate. If you want white papers, go to the professional section, broadcast and proaudio, technologies... or just go to the technical library:

www.dolby.com/technicallibrary.aspx?id=306&?TECHLIBITEM=8and

Quote:

---

And funny enough, on Sunday evening, I watched a DVD that had only a Dolby soundtrack and its dynamic range was clearly too great for my Sony Bravia flat screen TV: either the loud bits were too loud, or the quiet bits were too quiet, thus forcing me to adjust constantly the volume during playback, which was very annoying.

---

Sounds like the Dolby Digital DRC mode is switched off in your system at present.

Quote:

---

It sounds like the whole Dolby thing is mainly for playback in commercial cinemas and for commercial broadcast, and mainly a marketing ploy for other contexts, where it sometimes even results in worse audio, as was the case for me on Sunday evening.

---

Dolby goes to enormous lengths to try to educate the public, as a look around their website will surely have told you, and most DVD and home theatre handbooks are quite clear about the DRC functions.

In my experience most film projectionists are pretty professional, but in any case, the Dolby Digital metadata isn't employed in standard theatres -- it is a tool intended entirely for the domestic viewer.

The reasons you experienced 'worse audio' was simply because you failed to understand and confiigure the system you chose to use, and attempted to listen to cinema dynamics on a replay system that couldn't cope.

Hugh

--------------------
Technical Editor, Sound On Sound

Thanks for your additional comments, Hugh. Mine are below.

Robert

**********

Hugh Quote:

---

Usually there is a setting to enable Dolby's DRC somewhere in a set up menu, or the analogue outputs are preset to always apply dynamic range reduction.

---

I remember seeing that now. Thanks.

Hugh Quote:

---

Dialnorm is always applied and is essentially a remotely preset volume control (determined yet again by the dubbing mixer) that allows the dialogue from films and TV shows and other material on DVD to sound roughly the same loudness.

---

Sounds like Dialnorm is compression of a certain subrange of the total dynamic range, where that subrange is the range in which dialogue is typically found.

Quote Robert:

---

I hunted around that site for about 15 minutes and found mainly marketing information, no white papers, not a single mention of compression, etc. Can you indicate a particular page on that site that has sufficiently detailed technical information?

---

Quote Hugh:

---

Come on Robert. It's a superbly designed site and very easy to navigate.

---

I don't disagree with that, but after following your advice here and looking at the site a bit more, it's only reinforced my belief that the various technologies and DVD player settings are just too numerous for most folks to bother with. For starters, there are three or four variations of "Dolby Digital", and who, other than specialists, such as you and me, is going to bother educating themselves about them?

Quote Hugh:

---

The viewer configures their DVD or home theatre system to either ignore the values for the full cinema experience, of apply them for a more restrained performance. The metadata simply records a value that determines one of many pre-set dynamic range reduction curves that the dubbing mixer decided was the most appropriate for the material.

---

And so what happens if you select "Dolby 5.1" and turn DRC on? Then do you get the soundtrack compressed twice? Where is the answer to this question? On that Web site somewhere? And who is going to insert a DVD into a DVD player, then see that these settings can to be adjusted, but then stop and go look at Dolby's Web site?

Quote Robert:

---

Dolby goes to enormous lengths to try to educate the public, as a look around their website will surely have told you, and most DVD and home theatre handbooks are quite clear about the DRC functions.

---

I don't know anyone who even skims through those handbooks, except me. Most people I know, including directors and producers, pay someone to come set things up, then say to the technician, "just tell me which button to push to play a DVD". It is simply inconceivable that any of the people would ever bother to select Dolby 5.1 or some other soundtrack, or bother to turn DRC on or off. And if these film festivals operate at all like the music festivals that I've had my works performed at, the person in charge of playback will not change a thing with the set-up in order to accommodate a particular audiovisual work. (At music festivals, no changes to fader positions on the mixing board, etc.) Perhaps this is why my composer-associate gave up a long time ago and simply goes to film festivals with two DVDs, one with compressed audio and the other without.

Quote Hugh:

---

In my experience most film projectionists are pretty professional...

---

I don't doubt that, but, based on what I've heard, at a film festival, it's just a ninny who will insert a DVD and press "play", changing nothing with the player's set-up, even if you beg him to do so.

As an anecdote, before manually confining the dynamic range of my soundtrack to 45 dB, I sent a particular scene to someone associated with "the business", whose opinion I respect, and who is an electrical engineer and very technically adept, and told him that, due to the wide dynamic range of the soundtrack, either he should play it on his super home theater system (which he indeed has), or play it on his computer but listen with headphones. Well, he just wouldn't do it. Instead, he insisted on playing it on his iMac, listening only via the computer's built-in speakers, then complained, saying, "I can't hear the quiet bits, even with the volume turned all the way up. I know you want as wide a dynamic range as possible, but you've got to compress the soundtrack." And I find that this kind of unwillingness to make any arrangements for better sound during playback is typical, most unfortunately.

As another anecdote, a long time ago, I was in a car with Hugh Padgham, who was driving, although I didn't know who he was at that particular moment. He asked me to pass a cassette tape to him that was laying next to me, on the back seat. As I did so, I noticed that something like, "Sting master version II" was written on the cassette insert. He inserted the tape into his car stereo, comprising a Nakamichi in-dash cassette deck with super amplifier in the boot and eight speakers, or whatever, then fooled around with the deck's head alignment adjustment, then started playback at deafening volume, then started to complain about inadequate something or another concerning the deck, although I couldn't hear at all the problem that he was on about, as things sounded great to me. I thought he was eccentric until later that evening, when I asked him what he did for a living... But my point is that someone like him will take the time to ensure sufficiently loud and in general, proper playback of a musical work, and probably an audiovisual work too, I imagine, but that's because he's a recording engineer. Most other folks will not -- not even those in the business.

Quote Hugh:

---

The reasons you experienced 'worse audio' was simply because you failed to understand and confiigure the system you chose to use...

---

I didn't "fail" at anything, because I didn't even try, since I didn't want to be bothered! By the way, I now remember turning DRC off, a few months ago, when I had read what it was, as I wanted to hear what the audio on my project would sound like on a TV without any compression applied . But I dare not turn DRC back on, because I may forget to turn it back off when checking once again the audio of my project!

Too many options, too many settings to worry about, all in the name of "new and improved", but all at too high a cost -- a cost that just doesn't make it worthwhile. That's my opinion, anyway.

--------------------
Robert Eidschun

Quote rhd_toyl:

---

Thanks for explaining that - I thought two's complement binary was a number system where the msb represented the polarity of the following bits. I'm actually starting to regret weighing in on something I don't know very much about. . .

---

Do you mean the parity of the following bits?

Anyway, I only know of the application of twos complement arithmetic to multiplication, as I explained it in my previous posting.

--------------------
Robert Eidschun

Hugh,

I just read through the chapter on Dolby encoding in the manual for Compressor, which is Apple's utility for preparing video (with audio tracks) for DVD, and what I came across is disconcerting, given what you've explained:

1) Normal dialogue should be at -31 dBFS, which is below the bottom end of the range that you mentioned. So, when setting up DialNorm, you specify the level of your dialogue, say, -15 dbFS, and Compressor will then attenuate your entire audio track by 16 dB.

2) Compression should be applied for playback in cinemas, but rarely for playback in any other context(!).

Here is a link to the manual, so that you can read for yourself:

http://eidschun.fastmail.fm/Compressor_User_Manual.pdf

--------------------
Robert Eidschun

Quote Robert Eidschun:

---

Sounds like Dialnorm is compression of a certain subrange of the total dynamic range, where that subrange is the range in which dialogue is typically found.

---

Why use three words when 25 will do? No, Dialnorm is, as I said, a remote static volume control. It provides fixed level offset for the entire programme. The reference point is the nominal volume of dialogue, and an offset value is entered into the metadata such that the dialogue is replayed at the same nominal level for all programmes. Essentially, the volume is adjusted on replay such that the dialogue is replayed at a nominal level of -31dBFS. Again, lots on this in Dolby's site.

Quote:

---

For starters, there are three or four variations of "Dolby Digital"

---

Not really. Dolby Digital is universally found on DVDs and is used by most HD broadcasters as the delivery format. it is the system most people will have some experience of. Dolby Digital EX is a marketing name but doesn't change much. It is exactly the same as Dolby Digital, but with a phantom centre rear channel added using a simple matrix process. You can achieve exactly the same thing by routing a source equally to the two rear channels. The decoder simply recognises phantom rear images and routes them to an extra rear centre speaker.

Dolby Digital Plus is an updated version of DD with additional facilities made available to suit the greater capabilities of BlueRay and HD broadcasts. However, because it is relatively new and there is not that much hardware capable of supporting DD+ out on the market yet, few if any broadcasters are using it, and most home applications are using 'legacy' DD decoders.

Dolby True HD is an entirely different system which is loss-less and isn't relevant to broadcasting.

Quote:

---

who, other than specialists, such as you and me, is going to bother educating themselves about them?

---

You'd be surprised. There is a rapidly growing market for home theatre using some extremely sophisticated equipment and some remarkably well informed users. Just look at the growing number of magazines that cover this area on the newsagent shelves.

Quote:

---

And so what happens if you select "Dolby 5.1" and turn DRC on? Then do you get the soundtrack compressed twice?

---

No. I think you are confused over the meaning of 'compression' which is different in different contexts.

Dolby Digital uses data reduction (often erroneously called data compression) as a means of making the file sizes more manageable for storage or transmission. But this has nothing to do with dynamic range compression.

Dolby Digital soundtracks are always full range -- they have the original wide dynanmic range as mixed at source -- and in the case of cinema films, the material has the dynanmic range intended for theatre reproduction.

The metadata carried with Dolby Digital soundtracks has encoded information which states which of the several different dynamic range compressions curves that are available to impose if the user seleects DRC mode.

There is only one DRC process involved.

Selecting a stereo PCM track instead of the Dolby accesses an entirely different mix which may or may not have been created with a smaller dynamic range.

Quote:

---

Where is the answer to this question? On that Web site somewhere?

---

yes...

Quote:

---

And who is going to insert a DVD into a DVD player, then see that these settings can to be adjusted, but then stop and go look at Dolby's Web site?

---

Normally, the selection of whether or not to use DRC is part of the installation process -- a one-off setting for most people that is performed at the same time as choosing the video format and everything else -- and the DVD player handbook normally explainns the ideas fairly well. No need for most people to trawl the Dobley site.

Quote:

---

It is simply inconceivable that any of the people would ever bother to select Dolby 5.1 or some other soundtrack, or bother to turn DRC on or off.

---

We obviosuly move in different circles. In the UK there is some consumer ignorance about this, of course, but most people who have a serious interest in high quality sound and DVD are very knowledeagable -- and the consumer magazines explain it all quite regularly.

Quote:

---

And if these film festivals operate at all like the music festivals that I've had my works performed at, the person in charge of playback will not change a thing with the set-up in order to accommodate a particular audiovisual work. (At music festivals, no changes to fader positions on the mixing board, etc.)

---

Clearly, that is just plain unprofessional... but perhaps not entirely unusual, sadly.

Quote:

---

Too many options, too many settings to worry about, all in the name of "new and improved", but all at too high a cost -- a cost that just doesn't make it worthwhile. That's my opinion, anyway.

---

Sounds a bit like this new fangled music with too many notes to me...

The system is what it is, and in many respects it offers considerable advantages over the preceding technologies. But as with anything new, it requires a degree of effort on the part of the user to comprehend and make best use of what is on offer.

Hopefuly, this discussion and the other information I have suggested you research will help in that process.

hugh

--------------------
Technical Editor, Sound On Sound

Quote Robert Eidschun:

---

I just read through the chapter on Dolby encoding in the manual for Compressor, which is Apple's utility for preparing video (with audio tracks) for DVD, and what I came across is disconcerting, given what you've explained:

---

Robert, authoring metadata for Dolby Digital is a non-trivial thing. I regularly teach it to professionals and it takes two days for the full thing.

Can I suggest you do yourself a huge favour and either spend the time and effort reading through the full set of Dolby's papers on Dolby Digital and its metadata, or just ignore it completely. This is not an area to poke at from a distance with a long stick while complaining about it being too complicated with too many options.

Quote:

---

1) Normal dialogue should be at -31 dBFS, which is below the bottom end of the range that you mentioned. So, when setting up DialNorm, you specify the level of your dialogue, say, -15 dbFS, and Compressor will then attenuate your entire audio track by 16 dB.

---

The replay level is adjusted by the dialnorm metadata to set the nominal dialogue to -31dBFS. That is entirely separate from the industry conventions for the source recording.

And there is no compression involved. Dialnorm is a static, fixed level offset.

Quote:

---

2) Compression should be applied for playback in cinemas, but rarely for playback in any other context(!).

---

Dialnorm is usually applied in all Dolby Digital replay situations, cinema and domestic alike. Dynamic range compression is never applied in cinemas, but can be applied in other situations by the user.

The thresholds of the preset DRC curves are related to the dialnorm adjusted replay level at -31dBFS -- so it is critical that the dialnorm value is sete correctly in order that the DRC functions as intended.

Quote:

---

Here is a link to the manual, so that you can read for yourself:

---

Thanks. I think I'm happy to rely on Dolby's own information -- both as presented on its website and as taught to me first hand to enable me to teach it to audio professionals on their behalf -- rather than on the potential misunderstandings and confusions of a third party developer.

Hugh

--------------------
Technical Editor, Sound On Sound

Quote rhd_toyl:

---

Thanks for explaining that - I thought two's complement binary was a number system where the msb represented the polarity of the following bits. I'm actually starting to regret weighing in on something I don't know very much about. . .

---

Don't regret it -- you are quite correct. Two's Complement is a binary counting system in which the upper half of the binary number range has been redefined to represent negative quantities. It is used almost universally as a coding scheme for audio signals in converters, DSPs and DAWs.

John Watkinson explains all this in chapter 3 of his reference book (the Art of Digital Audio).

There are several reasons for its use. First, it is easy to define a 'zero' point in the midrange. In a four bit system that would be a value of 0000.

In the vast majority of applications positive going signals use those binary values where the MSB is set to 0, while negative signals are those binary counts where the MSB is set to 1. Thus a full positive excursion would be represented by 0111 and a fully negative excursion by 1000

The other significant advantage that stems from this redefined centre zero is that it makes the mixing of audio signals very simple, as concurrent samples can simply be added using pure binary calculations to produce a result which is also referred to the midrange exactly as required.

Hugh

--------------------
Technical Editor, Sound On Sound

Quote Robert Eidschun:

---

Do you mean the parity of the following bits?

---

No, he didn't mean that, and he is quite correct. I think you may be straying into the territory here where a little knowledge starts to become dangerous

hugh

--------------------
Technical Editor, Sound On Sound