Βι¶ΉΤΌΕΔ HDTV: "The Βι¶ΉΤΌΕΔ's Bold Trial Of Reverse Karaoke!"
Post categories: HDTV
| 16:07 UK time, Thursday, 17 July 2008
I finished my last post with the sentence "Two days later the World Cup kicked off and then the fun really started...".
I wasn't joking. It's been an amazing two years, with many successes and a few disasters!
Top of the list for "must try harder", though, is audio - with surround sound causing more problems than anything else. I couldn't decide what to call this post - I started out with: "Surround Sound is easy... ...to get wrong!"
But then there was an incident with the Eurovision Song Contest and I had a different idea. So, welcome to:
The Βι¶ΉΤΌΕΔ's Bold Trial Of Reverse Karaoke!
(Or, We Send The Vocals; You Provide The Accompaniment)
I'm sure that, like me, you've been to at least one Eurovision Song Contest party where you either had to dress up as one of the qualifiers, take a dish or bottle of the country you've been allocated, or been given a song sheet so you can sing along. This year, due to a technical fault, we had the ideal opportunity to try something new, so for around half an hour we sent the vocals for everyone to gather round the telly and provide their own instrumental accompaniment! From the comments in the many , I gather not many people took the opportunity, oh! well.
A couple of weeks ago, I was speaking to the technical director of : he said that over 75% of their quality control failures are due to problems.
Then you realise that Discovery are at least five years further down the HD road than us and they don't do that many live programmes! No excuse for getting it wrong, but it does put it into perspective.
So why and how does it go wrong and why isn't everything in surround?
Audio is a lot more complex than video and, for live programmes, the complexity is multiplied many times over. When we do live HD transmissions, the HD video feed goes directly to the HD channel and a down converted signal is sent to send to the channel for SD transmission. The SD channel has stereo audio, but we don't mix it down from the surround audio (yet); it is actually a separate mix at the moment.
It's worth remembering that well over 95% of the audience of any live programme is watching in standard definition with stereo sound and the standard definition audience must get the best possible service. This means, however, that we've just multiplied the problems by two.
On the day of the Song Contest, I had been providing my usual daughter taxi service and arrived home around 20:15. It did take a few moments to realise that something was wrong and that there should be a least a bit of music to go with the vocals.
My AV amp was saying 1.1 and this was a bit of a surprise because the transmission chain is supposed to be locked to 5.1 at the moment. This is another area for forum discussion and the Βι¶ΉΤΌΕΔ has been the subject of much derision, accusations that we don't know what we're doing and "what would Dolby say?" statements. So before going on with the Song Contest story, I'll give a bit of an explanation.
The HD channel only transmits one audio stream, and this is encoded either as 2.0 for stereo or 5.1 for surround. When we transmit a surround sound programme, the set-top box does a down-mix to stereo for anyone who listens via the analogue outputs or via the TV speakers. We send metadata in the audio stream to control the mixdown. It's worth remembering the metadata as they will come back to haunt us later!
When the trial service launched, we only had a few surround sound programmes - but I noticed, as we switched between 2.0 and 5.1, there were a few clicks on the audio when I listened at home. We tracked these for a while to try and find out what was happening, but they weren't serious and I initially put it down to my cheap AV system. Also there were no comments from the audience and even was quiet. Anyway, we had bigger audio issues to deal with at the time.
As we started to get more surround sound programmes and were switching between live, pre-recorded, surround and stereo in every direction, the clicks began to get a bit more annoying and there were a few forum comments, so now I knew others were hearing them too. We started doing some investigation and I even recorded every junction on my home set-top box over a weekend - sad or what? I had to programme it all manually as the guys at Sky wouldn't do an upgrade to allow me to record every programme junction, but not the programmes - advertisers' dream option, I thought!
Connecting my box to a few home cinema systems produced some very interesting results. Every box did something different - the more expensive ones actually muted the output for 2-3 seconds over each mode change; this was something we just couldn't tolerate. At the time, we were talking to Dolby about the problem, trying to find out whether our switching technology was at fault, but it looked like this was a home cinema system issue.
has also done some investigation and there seems to be a problem with the way some home cinema systems switch between surround modes. It's very similar to the aspect ratio switching issues we had in the early days of widescreen, i.e. unpredictable and variable between manufacturer and model. Dolby kindly sent us a release that explains the issues in more detail and we are looking at possible solutions to the problem. However, in the meantime we are staying locked to 5.1 - well, that's what I thought until the Song Contest came along!
I said earlier, I was distracted by other audio problems. Domestic audio systems use something called Dolby Digital or Dolby D to receive and decode multi-channel audio. We use another Dolby format, Dolby E, to carry up to 8 audio tracks in the space of a stereo pair. It's a very useful compression system, especially for outside broadcasts where there's limited capacity on the link back to Βι¶ΉΤΌΕΔ Television Centre. The Dolby E stream also carries all the metadata the sound supervisor has added to control levels and the stereo down mix. We can pass this onto the Dolby D encoder and then on to the decoder in a Dolby home cinema system or set-top box.
Audio signals are usually embedded into the video stream now, and Dolby E uses virtually every available bit in a PCM stereo pair - sending any measuring devise straight into the red zone! Because of this, every device we use to either pass or process the signal has to be properly aligned, timed and to carry the signal as specified by the standards bodies.
During the first summer of the trial, we had been trying surround sound on different events - not only to get experience capturing, but also to try moving the signal around our infrastructure. Wimbledon had been a great success, and we were looking forward to the Last Night Of The Proms being the climax to a summer of live HD.
The way we get signals back to TVC is now critical not just for the quality of the video but also for the successful delivery of surround audio. During Wimbledon, we used a fibre link and, if you remember my last post, our issues were all about getting the fibre into the switching centre and were nothing to do with the quality of the signal it carried! To get signals back from the Royal Albert Hall, we used a microwave link to the PO Tower and then our permanent connections back to TVC. Not much of a problem - we've been doing it in SD for years and it only needs a few more Mbs for the video and an additional Dolby E stream for the surround sound. What we discovered during tests, though, was that the video and the stereo went in and popped out the other end with no problems but the Dolby stream just didn't go anywhere.
It was during these tests my programme guide dropped through the letterbox at home, and it was advertising (on the same night as Last Night Of The Proms, no less), a live HD concert by a beat combo led by someone called "Robbie Williams". This was to come from Manchester, if I remember correctly, and I would have laid money that it would be in surround and the whole package would come down to Osterley via a satellite link.
After a few phone calls and a couple of beers, I discover that just like us, the Dolby E stream was going into Sky's sat link and resolutely refusing to come out at the other end! Both of us had identified the broadcast MPEG2 encoder as the culprit but we were not getting anywhere with a solution. Needless to say, we pushed very hard and a few days before the Saturday night, an updated version of the firmware arrived and the surround burst out of the other end of the chain.
By now (I hope), you have forgotten what this post is all about! But just in case you still want an explanation about the Song Contest... At 20:20, I was on the phone to the duty engineers asking why I hadn't been told we were doing a karaoke trial! They were very well aware of what was going on and were desperately trying to find the problem, but they did put me through to the HD network director and we decided to go immediately to an up-converted Βι¶ΉΤΌΕΔ ONE feed - this gave us properly mixed stereo audio, but of course we lost the HD. I said I would call the engineers back in ten minutes to see how the fault-finding was going. [ by on Flickr]
Now, the next question you want to know is: "Why couldn't you take the HD pictures with stereo sound?". For that, you'll need to read on...
At 21:00, I was talking to the duty engineers who had traced the audio back through the chain, but could find nothing wrong. It is worth a bit of time to tell you how the Eurovision Song Contest was passed through the chain from the venue to you.
The UK is a voting member so we have to feed ourselves back to the venue for the voting section of the programme as well as mixing Terry into the main clean feed coming in. To do this, we use a studio gallery. The engineers had traced the audio all the way back to the studio and listened to the direct feed from the venue, and at every point it was all there and correctly mixed. Everyone at every monitoring point could hear perfectly mixed surround sound and because of this, everyone was convinced it was a problem with the Dolby-E-to-Dolby-D conversion - the very last process in the chain. Meanwhile, the studio team (the very first point in the chain) was sure that all the settings were correct and that they could decode the Dolby E properly to prove it - but the Dolby D encoders were telling us we were sending 1.1 and nothing we could say would convince them otherwise!
Time (and contestants) were passing fast, so it was time for a change of tack and I am not one to believe the obvious - especially when two devices are saying something different to everyone else!
Earlier, I asked you to remember metadata, and by now I was thinking metadata not audio. I went back through the chain, but this time asked about the metadata that were nothing to do with operations (eg mix-down settings, dial-norm, etc).
I was getting grief all through this, as my daughter had now decided the Song Contest was not on her viewing list for a Saturday night and how she was to call several boys she had promised to call back if I was on the phone and she had to use the landline as I had deliberately let her mobile credit run out!
Anyway, and despite this going on in the other ear, several areas reported an inconsistency in some of the set up metadata readings but were not sure if it was a fault or a misreading as the results were different when monitored in different areas.
Before we did anything else, I needed to know what the actual metadata settings were leaving TVC and I also need to check lip sync because each Dolby E process delays the audio by 1 frame and we have to compensate by delaying the stereo and video to match. We had been checking and changing so much, by now there was no guarantee even if we did correct the fault anything useful would come out at home!
is probably the second biggest issue we've had during the last two years. Each device or process that a signal passes through introduces some delay from a few lines or microseconds up to several frames. This is one of the reasons embedded audio is so useful but even embedders and de-embedders introduce delay.
From the comments on some of the forums, you would think we had people sitting back watching out-of-sync feeds and doing nothing about it. We do take sync very seriously and every sync complaint is always investigated. We have an off-air recording of the HD channel so that we can go back and check if we need to. But what is sync really?
A while ago, C4, five, ITV, Sky, Virgin and the Βι¶ΉΤΌΕΔ met and agreed how we would define sync. "Easy", you say - "lips and voice should match!". Balls should hit rackets; boots should hit footballs etc at the right time! Not quite as easy if one of your signals is encoded and it takes one frame to decode it to see whether it's sync. Also, in a chain that has known video and audio delays, do you add all these together and compensate at the front of the chain, or at the back? Or, as we decided, do you insist that all audio is sync with the video as it leaves each area, no matter how it's processed?
We define sync as "in sync encoded" for Dolby E signals, so if you cut an embedded stream on a frame boundary, the Dolby E and the PCM stereo will cut in exactly the same place in the dialogue. Each area is responsible for correctly aligning signals to meet this requirement, including any material recorded to or played back from tape.
One other thing we've noticed is sync drifts in some devices depending on how hard the device is working. In a complex chain, there could be ten or more processing units and if they're all working flat out, a signal path that was in sync during line up can suddenly be noticeably out of sync. Also, it's worth noting that you probably have to include how hard the home set-top box is working too when looking at end to end AV sync issues!
The and other standards bodies are looking into sync specifications to see if the current standards are good enough for HD. We hope to see a tightening of the AV tolerance in all equipment soon.
Back to the Song Contest! To check exactly what our current state was, I asked the network director to switch back to the HD output so I could see what my AV amp did. That's why some of the forums reported a few seconds HD with mono audio! The metadata were still saying 1.1 on one of my boxes and 1.0 on another (please don't ask why I have two).
Now I had someone in the studio looking at the main Dolby E encoder, I was looking at a set up guide on the Dolby website (no, I don't have manuals at home) and I was talking through the settings I wanted to check. I really wanted them to check what the Dolby D metadata setting was, but no one could find it in the menu. Then one engineer asked if it could be the setting "AC3 metadata" - same thing I said, "What's it set to?". "Disable", was the reply - a quick change to "enable" and a phone chase back up the chain where everyone reported consistent metadata reading
So why did it ever get to air, what was wrong and why couldn't we do HD video with stereo audio?
Well, to start with, there was nothing wrong with the signal or the metadata and because of that the system did exactly what it as told to do - what was that? Simply put, we had told the system to ignore everything we told it to do! The embedded audio and metadata were not faulty, so the very simple system we put in place for the trial would not allow a non-standard change.
During the trial, we had to put in place a simple setup, but it was very well engineered to do a very simple job without intervention. Now we've been given permission for the HD Channel, we can spend the money needed to upgrade the system and to put in place all of the safeguards, monitoring and protection that a full channel requires. This has been going on around the "live" channel, and one of my first thoughts had been that some part of the upgrade was interfering with the main output.
The target for completion (including testing and acceptance) of the new infrastructure is the opening of the in August, which I gather is 8:08:08 pm on the 8/08/08! But we have just about got everything in place now, so from now on, we have the same type of transmission resilience in the chain as any other Βι¶ΉΤΌΕΔ channel. We also have the capacity to carry out the testing and trial we need to do whilst still on air.
It's my sincere hope that we are at the end of the first stage of our HD migration and that we can start to consolidate and to improve. So maybe this is not the sort of post you expected, but I hope it's given a bit more of the background of what we are up to and I would welcome comment or suggestions for what to do next.
Andy Quested is Principal Technologist, Βι¶ΉΤΌΕΔ Future Media & Technology
Comment number 1.
At 17th Jul 2008, Ed Lyons wrote:Very interesting blog. I'm unlikely to ever have to implement a surround sound system for a nascent HD channel, but you never know! It's good to see that there's interesting challenges in broadcasting technology :)
Looking forward to your future blogs!
Complain about this comment (Comment number 1)
Comment number 2.
At 17th Jul 2008, Joe Halliday wrote:Excellent post, great read. Thank you.
Complain about this comment (Comment number 2)
Comment number 3.
At 18th Jul 2008, Alan Robertson wrote:Very very thorough posting and really interesting to read about what's going on in the background - makes it amazing it works at all when you find out how complex it is!
Complain about this comment (Comment number 3)
Comment number 4.
At 18th Jul 2008, articulate wrote:Love what the Beeb are doing with the audio on their HD coverage of the British Golf Open
Complain about this comment (Comment number 4)
Comment number 5.
At 18th Jul 2008, Bill Taylor wrote:Thanks for the information.
DD 5.1 adds to the viewing experience so keep at it please!
Could you spell out how Βι¶ΉΤΌΕΔ uses the Sky information screen to describe the sound broadcast? I assume when you tag as DD this is the full 5.1 engineered sound dimensions to the program; where considerable resources have been deployed to give the best possible experience. DS means program is Stereo [I have noticed some programs with different sound from the rear channels, but not a proper balanced experience]. No DD or DS means?
Wimbledon sound was impressive, but variable. I enjoyed the crowd atmosphere, but at times they drowned out the commentators such that to give them the volume to discern them, I risked blasting the house with the crowd noise. I accept this is a difficult sound engineering problem, but believe it is worth the effort.
Thanks again.
Complain about this comment (Comment number 5)
Comment number 6.
At 18th Jul 2008, ShadowTD wrote:I've posted over on CableForum as opposed to DS, but I'd like to say this is the best blog post I've ever read on the internets.
I think the reason that people get so bent out of shape on forums is they are never convinced that anyone is actually doing anything about their concerns. Now I know if I ever see any issues with Βι¶ΉΤΌΕΔ HD that they are being investigated by someone who obviously cares very deeply about getting it right.
Being on Cable, Βι¶ΉΤΌΕΔ HD is the only channel I ever receive so it's very important to me. Keep up the good work and rest assured your efforts are appreciated by hordes of home cinema geeks.
Bonekickers looks especially good in HD on an 80" screen! :)
Complain about this comment (Comment number 6)
Comment number 7.
At 18th Jul 2008, Briantist wrote:It's nice to see the Βι¶ΉΤΌΕΔ investing time in proper sound. Apart from the Dolby Surround (analogue encoded) that you have carried in US programming in the past, I always thought the Βι¶ΉΤΌΕΔ ban on it (because of mono mixdown problems) a bit (ahem) short-sighted.
Shame that many people are reporting lipsync problems with the Freesat Foxsat HD boxes.
Anyway, thanks for the detailed posting.
Complain about this comment (Comment number 7)
Comment number 8.
At 18th Jul 2008, Digital Elysium wrote:Excellent post.
Complain about this comment (Comment number 8)
Comment number 9.
At 18th Jul 2008, MDL92 wrote:I'm biased as a Eurovision fan, but it's nice to see my favourite show is such a technical masterpiece.
Complain about this comment (Comment number 9)
Comment number 10.
At 19th Jul 2008, Dave Cheltenham wrote:Thanks Andy for your most interesting and informative Blog and for being so candid.
We expect problems on a new service and thought 2 years would have been enough to solve the obviously difficult interface problems and it is good to know that a full professional service will be fully operational from August.
One of the most aggravating things from all this is not from your department but presentation. The song contest is a good example in that through all the severe trials and tribulations, no on screen messages, no announcement, no apology, no nothing leaving the impression that the Βι¶ΉΤΌΕΔ could not care less and the viewer will take what they are given. They partly redeemed themselves during the Euro 2008 thunderstorm break.
The 2008 Proms HD audio started true to form but fortunately this did not extend into the music. It was also good to see no HD Logo for this as well. Belt and braces, because I was out, I recorded the Βι¶ΉΤΌΕΔ2 as well as the HD broadcast.
Thanks again for great Blog - Regards
Complain about this comment (Comment number 10)
Comment number 11.
At 23rd Jul 2008, juhauk wrote:Thanks for the very elaborate post. Now my follow-up question is then who might have been in charge of the decision not to air the two semi-finals from Belgrade on the Tuesday and the Thursday that week, one of which the UK was also voting in? These two transmissions would have helped to iron out any technical problems before the main night that was Saturday.
Instead the Βι¶ΉΤΌΕΔ HD channel showed the Chelsea Flowers show. It is not like the channel would be overflowing with new HD material and supposedly the HD feed would have been available to the Βι¶ΉΤΌΕΔ with no extra cost either on the two semi nights.
Please take them all next year, and maybe by the third night it will work.
Complain about this comment (Comment number 11)
Comment number 12.
At 3rd Aug 2008, Darren Clift wrote:Really interesting post especially as I have noticed a problem with Last Choir Standing.
My av amp is working fine (recently upgraded to latest firmware by authorised dealership) and I can listen to all other Βι¶ΉΤΌΕΔ HD programmes but for the second week running the sound output has been zero on this programme only, although interestingly enough the amp shows a 2 channel signal detected.If I flick to bbc 1 sound is there, flick back and dolby 2.0 channel stereo is shown but no actual sound however high i increase the volume.
My AV amp is a Onkyo TX-SR875
Any thoughts would be gratefully received.
Complain about this comment (Comment number 12)
Comment number 13.
At 3rd Aug 2008, Andy Quested wrote:Dear Dazza124
I will check our metadata setting tomorrow but there have been no other complaints so I think I will have to point you back to the firmware upgrade.
Complain about this comment (Comment number 13)
Comment number 14.
At 3rd Aug 2008, Darren Clift wrote:Thanks for a quick response.
I am waiting/hoping that is a metadata problem as its making me wonder if something is wrong with my setup.
As previously mentioned all channels audio/video are fine.
I'm a avid Βι¶ΉΤΌΕΔ HD user as its the only HD channel I have access to but also because its got great programmes on it ( Heroes,Wild China, Six Nations, Wimbledon,etc,etc) all of which seem perfectly fine.
My provider is Virginmedia and I have their V+ PVR set top box.
I have a Bluray player aswell as the above all of which do not/have ever had this issue.
I cannot comment on the Eurovision issue as I never watched it.
On a seperate issue how likely is it we would have DD 5.1 on say the six nations rugby event (live events)?.
Oh by the way keep up the good work with the blog as it very interesting and talked about elsewhere on other dedicated HD blogs i visit.
Complain about this comment (Comment number 14)
Comment number 15.
At 4th Aug 2008, Andy Quested wrote:Dear Dazza124
It seems to be an issue with Virgin but we are looking into it. There were several complaints to Virgin so it would seem your equipment is OK. It seems the audio started again about 20mins in though.
Andy
Complain about this comment (Comment number 15)
Comment number 16.
At 4th Aug 2008, Andy Quested wrote:Sorry - far too many "seems" in that reply
Complain about this comment (Comment number 16)
Comment number 17.
At 5th Aug 2008, Peter wrote:Thanks for such a verbose post. I'm hoping this story is going to make it into the next series of Spooks!
Seriously though, very interesting to get an idea of everything that goes into a live HD broadcast.
Complain about this comment (Comment number 17)
Comment number 18.
At 9th Aug 2008, Paul1953 wrote:Very Interesting, a few more blogs like this about the p[roblems encountered with the move to HD would be much appreciated.
It shouild help the viewers to be more understanding when you have problems
Complain about this comment (Comment number 18)
Comment number 19.
At 10th Sep 2008, rg109xd wrote:Thanks for the illuminating blog, Andy. Since the Olympic deadline has now passed, are you happy that the lip sync problem is resiliently cured?
I ask because as a recent subscriber to SkyHD I have found several HD channels to have intolerable lip sync issues. The Βι¶ΉΤΌΕΔ HD channel seems ok but is that because you have taken Dolby 5.1 out of the equation? Or perhaps I just haven't noticed the transmissions in 5.1...
Certainly the main problems I do see are with 5.1 transmissions especially on Channel 4 HD, Sky Arts HD and Discovery HD.
Like many others I have read about, I'm considering downgrading until whoever is responsible sorts out the HD channel audio issues.
Complain about this comment (Comment number 19)
Comment number 20.
At 13th Jun 2009, U14033173 wrote:This comment was removed because the moderators found it broke the house rules. Explain.
Complain about this comment (Comment number 20)