quality Control
Automation
(Continued from previous Page) While it’s true this may be compromising the quality to some extent, it’s our job to push the technology forward to the point where viewers aren’t going to see a material difference, even with a 25-30% reduction in bandwidth. So, from today’s practical standpoint, H.264 hasn’t replaced MPEG2 and certainly H.265 isn’t going to replace either of them. While everyone is looking for the bandwidth savings with the newer codecs, I don’t see H.265 having a significant impact on the conventional terrestrial satellite infrastructure, which, as we've said before is still largely MPEG2 based. It’s mainly going to be an OTT technology, at least initially. I don’t want to appear to discount H.264 and H.265 -- it’s all part of the package -- but the larger message is not to fall into the trap of “now that H.265 is here, everything else is obsolete.” Look at what’s going on with MPEG-Dash, all MP4 based, so these things are additive. If you are trying to squeeze 4K content down an IP pipe, it’s a great solution, but there’s still lots going on and lots of bandwidth yet to be had in a technology that’s 15 years old.
QC in the a file-based multi-platform world
Content creators are constantly exploring new ways in which automation can improve efficiency. In many cases, the functions that they are automating aren’t necessarily complicated, but the reduction in mistakes and required manpower --particularly as the number of distribution channels to be serviced goes up-- is significant. It’s not as if all of those distribution channels are revenue generators either. In many cases, you have to supply content distributed through new mechanisms purely to appeal to customer/consumer convenience. At the same time, you don’t want the quality of your brand to be diminished by mistakes such as out of sync audio and bad captions that will disengage viewers. The quality of broadcast TV has declined somewhat over the years, and with so much content for viewers to choose from, broadcasters can’t afford to lose viewers on any platform because of quality issues.
The more versions of programs that have to be made --thanks in part to the multitude of platforms everyone has to deliver on-- the more difficult QC is becoming. Many issues can be relatively simple things like A/V sync, and caption quality or sync. So, we have been asking ourselves how could we better automate the process of QC, using more advanced methods such as machine vision techniques to limit the amount of manual intervention. The more versions you make, the higher the probability for error in those outputs. With so many content variants, every time you process it (even for tasks as simple as putting in spots or bumper trailers) and every time you touch the timeline, you run the risk of throwing the audio out of sync or losing captions and otherwise affecting the quality.
Automating the process of detecting and correcting, if possible, is something our customers would really like to see. At minimum, automatically detecting and diagnosing errors, and getting enough information to flag them is a key component to reducing the amount of human eyeballs required. A/V sync is something we’re particularly focused on with ways to automate the process of detection, setting thresholds and taking associated actions. I’m happy to report that Telestream is working on some exciting new developments in this area.
When discussing file-based workflows, people have a tendency to reduce the topic down to simply a case of transcoding. While encoding and transcoding are certainly a big part of it, when we strive for greater efficiencies, automated QC, measurement and analysis can have a significant impact. Anytime you process media there’s a potential to introduce errors, so you need your system to have enough intelligence to check the work, make informed choices and report out something like the following: “This process introduced artifacts at these 6 points in the program. They have been measured and found to be below the threshold of visibility” – while leaving the ultimate decision to manually QC the material or not up to you. So, as hot topics go, I believe automating QC measurement and analysis should be an important part of file-based multi-platform workflow considerations going forward.
43
Broadcast Beat Magazine / Sep-Dec, 2014