Radio 3 Subjective Technical Quality

[OldTechie]

Am I fooling myself? It seems to me that the technical quality of Radio 3 has improved since the BH refurbishment.

I recall hearing CDs that I knew well being played on Radio 3 and immediately deciding they were so wrong that I played them myself. I have not had that effect for a year or so.

Has anyone else noticed any improvement? If there is one, I wonder why? I think it was an all-digital installation in the old BH.

[johnb]

How are you listening? FM, DAB, Freeview, iPlayer?

[Gordon]

Don't know OT - BH has been all digital for a long time as have the inter-studio and distribution feeds to transmitters and other networks. I suspect that the new BH has more of the same. Can't say much re the quality of R3 because I listen mostly via DAB [better reception here] on a Sony "music centre" whilst I am working and so not atttending to the sound that much. I have not noticed any significant changes on the main system when listening to Proms - both FM and DAB - but then that's different from a normal programme with CDs. Perhaps they've just changed to new CD players?.

[Resurrection Man]

... Perhaps they've just changed to new CD players?.

Probably gone back to vinyl ........

[An_Inspector_Calls]

Don't know about any BH effect but listening via Freeview I'd say the (mp2) mid range distortion we used to hear is much less noticeable these days. Have they improved the encoders?

[jayne lee wilson]

Part of the problem with DAB and Freeview is the inconsistent, potentially misleading way these inadequate lossy codecs deal with different musical styles, acoustics etc. A striking feature of listening to the 320 kbps AAC webstream is how consistent it is in the sonic presentation from a given venue. It's easier to hear performance or production differences within that perceptual frame. Well-received FM would be good & consistent too, if it weren't for the hefty dynamic compression.

[OldTechie]

I should have said. My opinion that it is improved is based on 320K AAC listening.

My 192K MP2 comes via Virgin Media, but I guess it's the same as DAB as far as encoding is concerned (except when DAB drops to 160K.) I don't use Freeview because my TV digital output randomly gets left and right swapped and requires repowering to fix it. We don't have a respectable DAB receiver (the two portables we have are OK for gardening programmes.) It's not bad in my opinion. I sometimes use it when busy doing other things on the computer. I think I am finding this to be cleaner these days as well.

My FM reception is a bit odd. The receiver is OK, but the feed to the transmitter is peculiar. It was rebroadcast from Wrotham, but the BBC could not receive a clean signal for Radio 3 because of interference from a BT transmitter on the site. (It only broke Radio 3 - all others were fine.) In the early 90's an analogue line was installed (maybe at BT's expense) and it is still there. It suffers with the usual Optimod dynamic compression but also from hum because they ran the cable in through the power duct. A couple of years ago I had a moan at them and Arqiva spent two weeks half fixing it (and plugging up a temporary feed with left and right out of phase, requiring me to take a 10 minute walk to the transmitter hut and tell them.) Now the warm weather has come (and the last BBC analogue lines man probably left many years ago) the equalisation has gone off a bit - I reckon it's about 2dB down at 10KHz (measured by filtering the 320K feed until it sort-of matches by ear.) I guess that will go right again when the ground temperature drops back to normal. I'm pretty sure I thought it was OK six months ago.

I am guessing that the BBC is running internally at the EBU levels and is still using its standard PPM for monitoring them. That would put the 0dB line-up tone level at -18dBFS - reading "4" on a PPM. The peak PPM level is then expected to read "6", 8dB above line up or -10dBFS. That leaves 10dB of headroom for going over level without distortion. That is essential because the PPM has a 10mSec attack time so it fails to read very short peaks. The way I listen to the 320k feed is through foobar2000 which has a peak level meter (but nothing with BBC PPM characteristics.) It is a digital instantaneous peak meter so it reads the truth (except it should really be upsampled to at least 192K if it was being used for real control.) From the levels I see, I think the 320K feed is generated directly at the EBU levels. The 192K feeds seem to me (by ear) to be exactly the same level. I think the BBC is trying its best to deliver the digital feeds with minimum loss from their internal digital systems.

If there were no Optimod, the BBC PPM reading of 6 would correspond to 100% modulation on the transmitter. Transmitters had to have a hard limiter to cope with over levels from three causes: the operator might allow the signal to peak more than 6 on the PPM; the PPM might have failed to display the true peak; finally, there is HF pre-emphasis applied on FM transmissions, but there in no matching pre-emphasis on the metering. The Optimods will apply this limiting (including being a bit intelligent about handling the pre-emphasis) as well as adding dynamic compression (which does at least help to overcome the hum added at my local transmitter by Aqiva/BBC/BT.)

Last night, during the WOMAD stuff, there was a piece containing lots of spaced very loud hand claps. They were the key feature of the piece. They peaked -2dBFS on the foobar2000 meter. Normally the Optimod dynamic compression just ruins the dynamics of the music. However, these high level claps would have hit the Optimod's fast hard limiter. The key loud claps disappeared to background level on FM completely ruining the balance of the passage. The hard limiting of the 8dB over level plus the 2db HF loss in the feed to my local transmitter would have caused the claps to be attenuated 10dB relative to the sounds between them even if there was no Optimod. However, I think the Optimod's dynamic compression had raised the level by 6dB or so. The fast hard limiter had to chop off perhaps 14dB. It's the first time have ever found anything where the FM sound was hardly recognisable as being the same piece of music as the digital feeds. It usually sounds much the same except for being a slightly sweeter sound, with lost dynamic range, with added hum and, just for now, lost HF.

A-I-C - I think you may be hearing the improvement I am hearing - and because it is on both 192K and 320K feeds I think it may be a cleaned up BH system. All compression systems work better when given clean input because they don't waste bits encoding unwanted information.

[An_Inspector_Calls]

OldTechie,

Thanks for the fascinating reply. We really need a 'leak' to discover what's going on.

Do you have any idea what sorts of noise the old BH system had on it?

I agree about the 192k/320k feed levels being the same, to the extent that if I'm recording a 320k feed and I suffer a drop-out (rare these days as well) I'll just splice-in a bit of 192k if I have it to hand. I'd prefer the 320k feed all the time, but would be penalised by download limits. Perhaps now DTT is about to get a chunk of 600 MHz spectrum, allowing ten new HD channels, we might get a few more buts as a spin-off?

[aka Calum Da Jazbo]

R3 is sounding increasingly better in Uruguay

[ferneyhoughgeliebte]

R3 is sounding increasingly better in Uruguay

- more mellow?

[Gordon]

R3 is sounding increasingly better in Uruguay

Distance ends enchantment perhaps?

Perhaps now DTT is about to get a chunk of 600 MHz spectrum, allowing ten new HD channels, we might get a few more buts as a spin-off?

#8 Beware euphoria.
(a)The HD channels, up next year, will be HD versions of existing content, not new, for the time being,

(b) the licence conditions are until 2026 BUT they allow the government to revoke them at notice with the likelihood being in 2018 just when consumers have got around to getting the right receivers to get these HD services,

(c) the mobile phone companies are livid that this has happened because they want the 700 MHz spectrum for new 4G services and so want the existing DTT channels all to be pushed into this 600 MHz band. When the Europe wide 700 MHz agreement happens [expected in 2015] the spectrum for DTT will be reduced and packed into the 600 MHz band; that means a number of things possibly the dropping of these HD channels that are bit rate hungry.

[An_Inspector_Calls]

I wasn't expecting new content.

I've been watching 4G consume the DTT band over the last few months, through both the need to retune several times and from e-mail postings from Ofcom. So I was quite surprised about the 600 MHz allocation. I've also been surprised at how much bandwidth 4G was consuming; greedy isn't it? Anyone working on 1024 QAM? Perhaps we should all switch to satellite TV?

[Gordon]

I wasn't expecting new content.

I've been watching 4G consume the DTT band over the last few months, through both the need to retune several times and from e-mail postings from Ofcom. So I was quite surprised about the 600 MHz allocation. I've also been surprised at how much bandwidth 4G was consuming; greedy isn't it? Anyone working on 1024 QAM? Perhaps we should all switch to satellite TV?

Greedy it sure is. The original UHF frequency band [470 – 862 MHz] was allocated to analogue television in 1961; the individual channels, numbered 21 to 69, were each 8 MHz wide and grouped as shown below. As a result of DSO, which completed at the end of 2012, this UHF spectrum was replanned such that the Lower 600 and 800 MHz sub-bands were recovered. The Lower 600 MHz sub-band was reserved but the 800 has already been allocated to 4G mobile communications and the networks are currently being built.



The sub-bands and the channel/bandwidth allocations are as shown in this Table:



This currently leaves DTT with 32 channels to support 6 national service multiplexes, ie 5.1 channels /multiplex, not all of which can achieve full national coverage. All DTT channels are available for White Space use where locally feasible. All these DTT channels, including both 600 MHz sub-bands, have been co-ordinated for DTT with appropriate European neighbours.

This new award by OfCom gives the Lower 600 MHz sub-band to DTT for 2 more multiplexes to carry “new” national HDTV services in 7 channels, ie 3.5 channels/multiplex; altogether then this allows 8 multiplexes in 39 channels, ie 4.9 channels/multiplex. These may not all achieve full national coverage. This licence has a limited period such that the sub-band could be replanned if there was European agreement on the use of the 700 sub-band for mobiles in 2015. This could become effective sometime around or after 2018; if so, there would then be only 27 channels to support these 8 multiplexes, ie 3.5 channels/multiplex.

If the 700 MHz sub-band is allocated to mobiles another large retuning and interference mitigation process, similar to the one deployed for the 800 MHz sub-band, will be required.

One of the features of DVB-T2 is a more efficient modulation option BUT this comes at the price of increased Carrier/Noise or reduction in coverage. Off the top of my head I don't think 1024 QAM is in there, I think that the max is 256, 2 bits per symbol better than T1. The hope is that all the old SDTV streams based on MPEG2/DVB-T1 can be flushed out to leave only MPEG4/DVB-T2 and then there might be enough spectrum to carry them all. That flushing could take many years with simulcasting consuming scarce spectrum. For HDTV MPEG-4 and DVB-T2 is required; some existing DTT receivers bought between 2007 and 2012 may not have this ability. Getting all that done before 2018 could be tricky.

[Gordon]

A little bird tells me that there have been no recent changes to the technical infrastructure at BH or in the network. A few years ago they did change the coders, I remember seeing them during a visit about that time, but I forget exactly when, and we did have a bit of a board session on it at the time. It's possible you were fooling yourself.

[OldTechie]

A little bird tells me that there have been no recent changes to the technical infrastructure at BH or in the network. A few years ago they did change the coders, I remember seeing them during a visit about that time, but I forget exactly when, and we did have a bit of a board session on it at the time. It's possible you were fooling yourself.

Could well be...

Searching through a few things I came across this blog item:

http://www.bbc.co.uk/blogs/researchanddevelopment/2010/09/bbc-proms-in-extra-high-quality-n-the-internet--the-tech.shtml

This explains that the proms were fed into BH at 48KHz sampling rate, down-sampled to 44.1 for the continuity suite, up-sampled to 48kHz for the MP2 encoding area, and finally down-sampled to 44.1Khz for the AAC feed. The continuity suite was stuck at 44.1kHz because of the clip storage system. It says that was planned to be replaced in 2012 when they would change everything to 48kHz (excepts, of course, legacy content and CDs.) Does your little bird happen to know whether that has happened?

A couple of SRC actions removed from the proms path might explain why they are seeming cleaner, though it would be pretty marginal. Even if that has happened, it would have no effect on replayed CDs - the number of SRC events would be the same.

I think my comment about CDs must just be that they have thrown away some of the old players or maybe got more into their online storage so that studio CD players just do not get used any more. If you look at some pro CD players (e.g. http://www.hhb.co.uk/product_detail.php?id=312) you find that along with storage of lots of cues, RS232 control etc. you get +/- 12% pitch control. Being able to tune a pre-recorded track to the studio piano may be very useful - but the internal processing to do that and still produce a 44.1 digital output may not be loss free. Furthermore, this particular device does not seem to have a reference clock input so I suspect it would get wired into an analogue input to a sound desk where it would be digitised again. A DAC that can handle +/- 12% sample rate on its input may have sub-optimal filtering. Certainly the CDs I found objectionable were seriously reduced quality, not something subtle.