DAB Report

[Gordon]

So where I wonder did I get the idea of 80 seconds?? mea culpa. I'm even more impressed with you now that must have taken ages!! It might also explain why the resolution is not so fine, but I'd have to ponder that. Fully understand why you kept the 3 samples short!!

PS: I see now. In Bryn's spectrum plots there is a box at the bottom with "01:20:07:713" in it. I stupidly assumed this was minutes and seconds, not hours and minutes. It might go some way to explaining some of the features I'm puzzled by.

[Nevalti]

Here is a link to a sequence of the the close of the performance in each of the three transmission modes. It should be obvious which is which:

http://we.tl/jwTX3gmtv8.........

Many thanks for your hard work Bryn. They make interesting listening with a couple of surprises.

Sample 1: Completely dead and dull sounding with no detail, very little spacial information and absolutely no 'life' or realism. I had zero desire to keep listening to the music. In a word - awful!

Sample 2: A significant improvement with instrument timbre/texture now becoming apparent. Spacial information is improved and the auditorium is beginning to have some life in it. There is however, no way that one could be fooled into thinking that you were listening to real live musicians. It does very much sound like a cleaned up recording.

Sample 3: Sadly marred by surprisingly high 'hiss' levels masking much of the low level detail. Despite that hiss, there is significantly more realism to the sound. Instruments now have a very realistic texture to their sound which you can hear without the slightest difficulty. Those background creaks suddenly have enough detail that you know with certainty that they are creaks and not just clicks. Spacial information is not brilliant and locating instruments in the 'picture' is vague but you can hear room reverberations that simple do not exist in the other samples. This gives a far better 'you are there' impression and the whole atmosphere of the music is far, far more as you would experience if you were there. Apart from the hiss, you could easily convince yourself that you were listening to real musicians - live. The dynamics are just about noticeably compressed but that certainly did not mar that passage for me. In fact it probably helped.


Surprise 1: Your hiss level is higher than I experience. Bracknell is surely a good signal area? Is your aerial up to scratch?

Surprise 2: I thought that the 320 kbps aac feed would have sounded better than that. I have previously only made the comparison live switching back and forth so maybe we have lost something in the recording process? I will try it myself one day if I am ever clever enough to rule out potential recorder differences.

I am of course assuming that the order is DAB/320/FM. If that is wrong, it will be an even bigger surprise.

[Dave2002]

I am of course assuming that the order is DAB/320/FM. If that is wrong, it will be an even bigger surprise.

That will be interesting!

The first section is clearly significantly quieter, so will appear to be less "dynamic". It needs a couple of steps (maybe 3-5!) on the volume control to sound similar to the other two. At first I suspected that the "noises off" were less apparent in section 1 (about 20, 24 and 41 secs in) - but with the volume raised they're probably similar. The outburst at the end makes me suspect that this is in fact the HD stream - with a greater dynamic range. That would tie in with the labelling on the file.

I'm afraid my hearing must be shot, as I couldn't hear much hiss on any of the sections, or perhaps I didn't turn the volume up high enough.
This is interesting for me, as it ties in with my experience of receiving a recording via another ex board member somewhat associated with Couperin.

As I mentioned earlier, I haven't managed to get low hiss levels on FM for over a decade. I think the last section was in fact the FM. There wasn't such a lot to choose between that and the DAB on this occasion.

I listened both via speakers, and also via headphones, and several times - but hopefully not at very high levels. Possibly I could improve my hearing by removing wax from my ears, in which case I may be able to hear more detail. I do this from time to time, and sometimes I'm surprised at the higher levels and greater detail.

I will try again later using different headphones, and a different route - noting that Mr GongGong has suggested that the audio output of some laptops is poor (probably not his word!) - so I'll route for the headphones via one or more DACs.

[Gordon]

I will try again later using different headphones, and a different route - noting that Mr GongGong has suggested that the audio output of some laptops is poor (probably not his word!) - so I'll route for the headphones via one or more DACs.

MrGG has a point. I listened on my computer too, among other things, but a desktop whose noise level is, surprisingly, extremely low. My laptop however is like sitting in a fat fryer. Anyway to business:

Hm, a few comments first: looking at the file name Bryn gave the clips I'd guess that the order is HDS_DAB_FM, but perhaps that is flimsy evidence. Only he knows!!!

It is pretty clear to me that FM is the last, the hiss gives it away, unfortunately that clue doesn't help with anonymity. Having said that, there are other clues in the waveforms. This is the unmodified waveform of the clips that I posted above, the amplitude scale is logarithmic, ie in dB.



What do we see? There is a clear difference in DR for a start, and from this there would be no prizes for guessing that the last clip is FM, there is clearly some gain riding going on [the action of the Optimod or something else]. Similarly the first clip has a lower overall volume level but shows signs of greater DR than the others [and also has more bandwidth if you look at the spectra on an earlier post above]. So we would perhaps now guess that the first is HDS.

Taking this sequence of clips and listening verbatim as it were is valid enough because this is how they are presented by their respective delivery channels but I think we should allow some gain for the first clip in order to "normalise" them at the peak level. But I also think that listening at varying volume levels will not provide the best valid comparison. The central section has a near constant dynamic level falling by diminuendo to bar 819 and pp in string bass and bass clarinet prior to the sudden large ff, [the score says], whose amplitude is a good marker to observe DR. It is obvious that FM has its DR quite severely limited. By turning up the wick during the quiet part in the FM case whilst letting the peaks largely alone you will not be listening fairly because your ears will also respond to those volume differences.

Let's look at some waveforms again, this time these; they show both a linear vertical scale





and a log one of exactly the same data, which is which should be obvious:



So, as is visible from the waveforms, I have adjusted the volumes of the quiet part in each case to be as close to the same as I can but then letting the peaks go where they will. In these plots there is a tiny bit of clipping in the R channel of the first sample which I think is acceptable for this evaluation. You will see from the linear plot that I have been able to get the levels more or less the same but the linear plot shows some interesting features which I shall come to shortly. The result of this on the FM plot shows that that final impact of the ff will be pulled back and the level correction is not quick enough, it comes out of almost silence. The DR has been more or less settled for over 3 minutes so even though there was probably a look ahead going on in the DR processing, it failed to catch the impending volume increase.

Looking more closely at that 3 minutes or so quiet section we see that the first 2 samples are very similar in profile. But look at the FM plot, its volume is gradually increasing. At the start the levels are much the same as the other 2 but towards the end the gain has obviously gone up by about 3dB. This is an observation FWIW and over 3 minutes I wonder if it is that noticeable. The other 2 show no sign of manipulation but it is clear that the DR of the 1st sample is greater than the 2nd but by an amount much less than for FM.

Now to listening:

My first impression in all 3 clips is that the orchestra is close mic'd, probably multi-mic'd too, as per Proms practice, but I don't know that. Was anyone there to tell us? It sounds as though we are on top of the players and any hall ambience is largely lost but this is the RFH and I have not been there lately.** You can certainly hear the intake of breath just before the final ff. Anyway, to my ears none of them deliver a sufficient sense of space, they're all rather dry and I can’t say that any of them is better than another. Listen to the coughs and rustles in the audience and there is little liveliness from the reverb to those transient sounds out in the hall away from the mics. Listening also for the back and front of the orchestra and there is very little to be had. The soundstage in this section isn’t bad but Mahler has to take some responsibility for that as does the orchestra layout. Is that string bass at bar 819 on the left?? Is that where it was in the hall? I know that Klemperer sometimes had the basses on the extreme left in that hall.

Listening to the unadjusted sequence I’m suspicious of the effect of the large volume differences in the central quiet stretch, it puts that sample right “in your face” and my reaction was to leap for the volume. But shock of the ff is not there. I estimate that to equalise the volume levels that the first sample has been lifted by about 2 dB and the FM attenuated by about 12. Listening to the volume adjusted sequence, as per the second set of waveforms, those differences virtually disappear and the volume levels sound very similar to me, as they ought. The FM hiss is still audible but now it’s not nearly as intrusive as on the verbatim sequence. And that final ff has no punch.

So on the strength if this I can’t tell you which is HDS and DAB but I think the last sample is FM. But other evidence suggests HDS then DAB then FM.

A final comment for discussion: yes, compression like MP2 removes something from the original at a low level that may or may not be inaudible. But what if something is added to the sound instead and that that something, at a similar low level, disturbs what is present at that level and thus confuses it? Which is worse, which upsets you most? Low level reverb may well get attenuated but it can also get lost in noise.

**PS have a listen to this

http://www.bbc.co.uk/programmes/b03q4wtx

and see if you find a better ambience. I listened last night and thought it was quite good sound. The suggestion from the speech introduction is that the hall is quite live, certainly more so than RFH - yes I know there was PA involved. Once the music starts though there evidence of close mic'ing here too. The applause between movements and the intros have more space too. Don Quixote - a splendidly witty piece - try sampling the quieter episodes where the cello has dialogue with soloists.

[johnb]

Just to illustrate the effect of Optimod on FM and a comparison with DAB, some years ago I captured the DAB/FM/FW/iP broadcasts of a recording (that I already owned) of Shostakovich 12. This was some years ago but I think it still illustrates a point.

To be able to compare the DAB and FM on the same basis I first normalised the DAB (peak 0dB) I then normalised the FM to the same RMS level as the DAB - thinking that this comparison would reflect the situation is one were listening to both examples at the same overall level. (The CD was very similar to the DAB broadcast.)

This shows all four movements of Shostakovich 8. (I slightly "zoomed in" the vertical range so the actual 0dB level is off the displays.)

[Gordon]

Thanks John, very interesting, that DR compression is rather severe. If Optimod didn't do that job someone would have to instead by riding the gain like wot they used to. But to do that at the desk at the RFH would queer the channels that don't compress DR. It would then put that person back at the FM feed out of continuity instead.

It could always be left alone of course, which would get the portable and car listeners up in arms again, but it would also put the quiet bits of the output of a modern desk in the noise +THD. There is a difference, often confused, between DR and SNR, they aren't quite the same thing, one needs a decent margin for the quietest bits above any background disturbance like noise + THD. IOW SNR = DR + Margin.

Having said that of course there is a limit to what DR is tolerable in a domestic listening space. One can have too much of a good thing. Moderation in all things, including one's opinions perhaps.

[Nevalti]

Having said that of course there is a limit to what DR is tolerable in a domestic listening space. One can have too much of a good thing.

Yes indeed, and of course that portable radio in the kitchen is going to be completely incapable of reproducing the DR of FM, let alone DAB. So why is it people claim that DAB was intended or suitable for such radios?

Good dynamics are essential to achieve anything like realism but a high dynamic range is almost irrelevant in a domestic environment and often very undesirable. The ear can happily compensate for compression but it certainly can't compensate if you have to keep the volume down too low to even hear the quiet bits.

[Dave2002]

Good dynamics are essential to achieve anything like realism but a high dynamic range is almost irrelevant in a domestic environment and often very undesirable. The ear can happily compensate for compression but it certainly can't compensate if you have to keep the volume down too low to even hear the quiet bits.

Yes, but with digital technology it should be possible to broadcast the full dynamic range and have the DR modified either automatically or by a carefully controlled steering channel. That way everyone should be more or less happy. DAB does allow different degrees of dynamic compression depending on the receiver, which should work for different listening environments and situations.
Ppersonally I'd rather have the full DR available, providing I can kick in a dynamic range compressor if needed. It's hard doing things the other way round because some numpty has decided I'm not allowed to have the full DR.

[johnb]

The ear can happily compensate for compression but it certainly can't compensate if you have to keep the volume down too low to even hear the quiet bits.

Well mine can't. My ears, or should I say "my brain" can compensate for poorer quality audio (we all do that when listening to early recordings) but it can't for the DRC on FM, where the climaxes hit a glass ceiling (even you know there is a climax because of the orchestral timbre) and where I can never find a volume setting where both the climaxes and the quiet passages both feel right. (This is much less of an issue if you listen at relatively quiet levels.)

[Gordon]

So why is it people claim that DAB was intended or suitable for such radios?

Who says that? It's actually the other way round. Most people prefer to buy portable radios and as a result DAB as used is adequate for the majority. You say yourself that even FM is good enough for most of those people, indeed that is why Optimod is used, to limit the DR for them. You are the one who wants the better performance so that it can be enjoyed ona decent system. Part of that is a suitable DR which FM isn't providing for that case.

Dave is right, at least digital systems could have the means to manage the DR so that people can choose it for themselves. Sadly no system I'm aware of implements it.

[johnb]

The Arcam DT81 DAB Tuner that I bought over 10 years ago had the facility to apply various levels of DR compression.

[Dave2002]

Dave is right, at least digital systems could have the means to manage the DR so that people can choose it for themselves. Sadly no system I'm aware of implements it.

My car FM radio has a few features. Firstly I believe it has a few levels of dynamic compression - I don't know the precise details, but it can do that. Secondly, it is also linked somehow to the engine speed, so that the volume rises as the engine speed rises. Subjectively this can keep the music or speech volume about the same, as our perception is often not absolute, but relevant to ambient noise level.

I do use these features on occasion.

Most sets for use in a domestic environment would not need an engine speed feature, though maybe could modify levels depending on ambient background noise - e.g whether there's a vacuum cleaner turned on!

Technically these things are possible.

[Nevalti]

So why is it people claim that DAB was intended or suitable for such radios?

Who says that? It's actually the other way round. Most people prefer to buy portable radios and as a result DAB as used is adequate for the majority.

Several people have said it but here are your own words - "Many professionals I know believe that DAB is good enough sound quality for the mass of the population, listening as they do predominantly on kitchen portables, not on expensive HiFi. DAB was not, is not and never will be an audiophile system." That appears to be a pretty clear statement 'that DAB was intended or suitable for such radios'. I do accept that my paraphrasing of your words may have been accidentally misleading.

You say yourself that even FM is good enough for most of those people....

I didn't actually say that (paraphrasing again:() but I certainly agree that FM on a portable can be entirely adequate and is more suitable than DAB but I would happily see FM replaced - by something better.

You are the one who wants the better performance so that it can be enjoyed ona decent system. Part of that is a suitable DR which FM isn't providing for that case.

This is where I seem to be at odds with some of the people here. A high DR is not achievable on 99.9% of portable radios nor on a high proportion of hifi systems. I can accept that some people may want a high DR in their lounge but I don't recall ever sharing a listening experience with someone who actually did.

With a close-mic's recording mastered to give a high DR, if you set the volume at a level where you can hear and appreciate the quietest parts of the music, the loudest parts can be uncomfortable or even painful. In real life, some instruments are far too loud for me in a room the size of my lounge. In a large hall those loud sounds are not as close and you can somehow still hear the quiet bits. I'm not sure why. In a smallish hall, a trumpet blast or a rim shot (for example), close up, may well send my fingers immediately to my ears. A full orchestra, even from my preferred row 'M' or thereabouts, can reach levels only just below my discomfort level. Am I really alone on this forum in all this? If I am that may explain a lot but in the real world everybody I have spoken to shares my experience - approximately. In a domestic environment it is often me that wants the volume up at a realistic level whilst others want it lower. Conversely, some acquaintances continue to sit near the Jazz band after I have given up and moved further away - because it is uncomfortable and I don't want to damage my hearing.

So, no. A high DR is not necessarily desirable. I thought we had agreed that earlier. The ability of the whole system to reproduce dynamics is however essential in creating an illusion of reality.

It may be worth adding that for CDs I try to avoid close mic'd recordings and much prefer a simple crossed pair. KISS really does apply.

Dave is right, at least digital systems could have the means to manage the DR so that people can choose it for themselves. Sadly no system I'm aware of implements it.

Yes, that does seem to be essential - assuming that it is achieved without spoiling the music in the way that much digital processing does.

A couple of equipment examples have now been given above but that essential compression feature is obviously not generally available. I have never heard the Arcam DAB tuner (£600).

[Gordon]

Several people have said it but here are your own words - "Many professionals I know believe that DAB is good enough sound quality for the mass of the population, listening as they do predominantly on kitchen portables, not on expensive HiFi. DAB was not, is not and never will be an audiophile system." That appears to be a pretty clear statement 'that DAB was intended or suitable for such radios'. I do accept that my paraphrasing of your words may have been accidentally misleading.

I didn't actually say that (paraphrasing again:() but I certainly agree that FM on a portable can be entirely adequate and is more suitable than DAB but I would happily see FM replaced - by something better.

I think we may be getting somewhere! The word I would dispute is “intended”. I agree that if FM is to be replaced it should be something that better caters for the range of listeners’ needs.

When any new system of broadcasting is introduced it has to serve the public [within economic and other limits such as spectrum] in general. We would like it to be able to provide for the needs of everyone, whether they be casual listeners to speech etc, in a noisy room or in a car or even an “audiophile”. It would be good if we could do that with one system design, without loading its costs and implementation unduly for one small part of the audience. The great white hope – the “intention” if you like - was that DAB might be such a system, catering for all according to their condition. In addition, being digital, the system could offer other services as well as audio.

Had it been implemented differently it might have been acceptable by all. They could have just about got a PCM 44.1/16 system into the space, but only 1 service per 1.5 MHz RF channel – not good spectrum use compared eg to FM [ca 200 kHz]. But, first, in the interests of spectrum efficiency, they crammed too many services into the multiplex capacity and then used, and continued to use, a compression scheme that was bit hungry. At the time it was standardised there wasn’t much else that was mature enough – MP3 etc came later as did AAC. – but by the time that DAB was first implemented in the UK AAC was visible if not actually mature and standardised. I can’t remember when DAB+ was standardised.

So, as it stands it does not provide for people with high audio standards, the equipment to deliver them and the ability to hear well. However it is “suitable” for use with less capable equipment, and for people whose listening is not intensive and demanding. All this provided that there is sufficient signal strength where you are - another early failing in the system – and also that receivers have sufficient sensitivity which too many early ones didn’t. Those two latter factors did not help the early cause of DAB. Whilst these have improved the state of MP2 has not despite some attempts to improve codecs.

This is where I seem to be at odds with some of the people here. A high DR is not achievable on 99.9% of portable radios nor on a high proportion of hifi systems. I can accept that some people may want a high DR in their lounge but I don't recall ever sharing a listening experience with someone who actually did.

I’m not sure you are at odds. Agreed that a high DR is not suitable for portables – that is one reason why Optimod is used. A broadcasting system should, however, endeavour to at least transmit a DR that can be acceptable to demanding listeners in their homes – but what is that? What is the happy medium? People have complained on these boards about Optimod so it’s not necessarily working well, despite good intentions perhaps.

A system that allows the receiver to adapt would be a good thing. DAB has a facility to allow that [DRC] and many FM receivers in cars have for many years used all sorts of processing to help manage the DR, as Dave and others have pointed out. Perhaps that particular need is being dealt with. Once upon a time it was sound supervisors that kept an ear on the sound and an eye on the meters and tweaked from time to time now it’s a sophisticated machine that does it but at the send end, once for everyone. It is not impossible to put a DR feature in receivers only or even have an end to end scheme, like Dolby, DBX etc, perhaps or something similar. DR management is then moved so that listeners can choose. We would expect that whatever is done it is not obtrusive and audible in its actions. The puzzle of it all is why DR control doesn’t seem to interest receiver manufacturers. On paper it’s just an AGC but its attack, hold and release profile as well as the degree of control, will need to be adjustable by the listener. Can you get the sophistication of the Optimod into a consumer product?

With a close-mic's recording mastered to give a high DR, if you set the volume at a level where you can hear and appreciate the quietest parts of the music, the loudest parts can be uncomfortable or even painful. In real life, some instruments are far too loud for me in a room the size of my lounge. In a large hall those loud sounds are not as close and you can somehow still hear the quiet bits. I'm not sure why. In a smallish hall, a trumpet blast or a rim shot (for example), close up, may well send my fingers immediately to my ears. A full orchestra, even from my preferred row 'M' or thereabouts, can reach levels only just below my discomfort level. Am I really alone on this forum in all this? If I am that may explain a lot but in the real world everybody I have spoken to shares my experience - approximately. In a domestic environment it is often me that wants the volume up at a realistic level whilst others want it lower. Conversely, some acquaintances continue to sit near the Jazz band after I have given up and moved further away - because it is uncomfortable and I don't want to damage my hearing.

So, no. A high DR is not necessarily desirable. I thought we had agreed that earlier. The ability of the whole system to reproduce dynamics is however essential in creating an illusion of reality.

It may be worth adding that for CDs I try to avoid close mic'd recordings and much prefer a simple crossed pair. KISS really does apply.

I agree or at least have sympathy with much of that. Concert goers have a degree of choice as to where they prefer to sit to avoid excessive DR in the real world. You don’t really want real world DR in your home. So how much should a broadcaster or CD maker send? Dilemma- pleasing all the people all the time. My tuppence is that FM hasn't enough and DAB is about right. Soluble by having DR adjustable in the home – so if it is important, and I think we seem to be agreeing that it is, why do we not do it?

I agree re KISS too!! One reason that late 50s early 60s stereo is prized is because they used simpler mic technique and only had few tracks to play with. As to Blumlein crossed pairs, EMI used it at first but graduated away; the BBC used it a lot in the early days too. If you used two figure of 8 mics in a crossed pair you automatically got ambience. Decca used a hybrid with omnis and cardioids. RCA and Mercury famously used spaced omnis and people seemed to like it. Who was right? Both it would seem, depending who’s listening.

PS: it occurs to me that broadband delivery in real time ie streaming and file transfers both ought to have easy means for DR control whereas broadcasting infrastructures do not for several reasons, some obvious some not. Do any of the iPlayers [includes Windows Media etc etc] offer a means of setting DR? As they are downloadable/upgradable .exe files why not? It might mean more processing power to enhance the decoder. I can't find such a means in any of the players on this machine = WM, iTunes, Quicktime and a few other proprietary programs.

If you import your downloaded file from iTunes or whoever, you can import it into say Audacity or one of the more fully functioned programs that you pay for, and set the DR to whatever you want, off line so as not to challenge computing speed, resave the file and listen to that keeping the original in archive. Perhaps an advantage for downloading?? The disadvantage is of course that some of these files are compressed and one would need to be happy with the compression, the decode software and any sample rate and bit depth changing software that one uses. With FLAC one is safe enough but have to accept larger file sizes - so what for a one time download - and perhaps m4a at a good bit rate.

If you find that your CDs are also giving excessive DR [eg for use in your car or on your kitchen player] then the same can be done - rip it to your computer, process the wav file as above, re-save and burn. A faff of course but it might solve a problem?