How Do You Get From Producer to Viewer?

Alert readers may recall that I started to open a new can of audio worms last month.

I began a series of columns that are to be devoted to the perils of upmixing and downmixing. This month, as promised, I will take a look at the various ways we can get into trouble while transmitting our beloved programming from the production suite to the home viewer.

Just so you know, I'm talking about the emerging world of ATSC transmission, not the traditional NTSC analog transmission chain. And while there are similarities, this new world has significantly increased complexity along with its increased potential.

I have to admit to some ignorance. I don't really know all the steps used to get the program into living rooms. So, I asked a Canadian for some help. Michael Nunan, as I mentioned in last month's column, is the post-sound supervisor for CTV in Toronto. He has been an avid and careful student of this, and can reasonably be considered to be an expert's expert. Thank you, Michael!

As you probably all know, programs are either produced by networks or sold to them, or sometimes produced for an independent such as the Discovery Channel. Network programs are transmitted to affiliate stations. From there, numerous things happen:

The programs are transmitted via terrestrial antenna to viewers (in various formats, including analog and digital HD ATSC formats); or they're distributed to cable and satellite distributors, who then distribute them to their subscribers, possibly with additional processing.

The independent channels transmit programs directly to the cable and satellite distributors.


This is where it gets tricky and weird. The "big four" networks all have their own ways of getting programming to their affiliates.

One network uses Dolby E for the audio (a digital format with comparatively mild and robust data compression designed for video transfer between broadcast facilities). Interestingly, sometimes metadata is included as part of the Dolby E digital transmission, while other times it is encoded with the video picture.

Another network transmits the program material audio encoded as AC-3 compressed data. The affiliate will decode it and, after adding their local materials, will re-encode it as AC-3 to be transmitted to the home viewers and other distributors. No metadata is carried with this, so it has to be included with the picture.

The other network transmits the audio as four PCM or MPEG-encoded stereo pairs, with metadata included in the picture.

It is a time-honored tradition in radio broadcasting for local stations to process the audio to create a distinctive "signature" sound. This usually consists of numerous stereo processors placed in line between the output of whatever console is being used and the transmitter.

As we've seen from above, "transmitter" is a somewhat loose term now, but the tradition remains. Signal processing will almost always include audio compression and sometimes EQ. Sometimes, stereo enhancement or "stereo synthesis" may be included as well.

This is processing that takes place on just left and right channels. It is not clear in any given case whether or not it takes place before encoding to AC-3 for the digital clients. It is also not clear, in any given case, whether or not such processing happens multiple times to a signal, as in having such processing done by the affiliate and then occasionally by the cable operator. According to Nunan, this happens fairly often.

If left and right are to be encoded as part of a multichannel AC-3 signal, or as part of a Dolby Pro Logic matrixed signal, then signal processing done only on the left and right channels can introduce significant errors for any subsequent upmix or downmix.

It gets worse. Surround sound comes in two flavors: discrete digital signal arrays (Dolby Digital) and matrixed analog arrays (Dolby Pro Logic). These are fundamentally and profoundly different, and need to be approached in different ways.

Where it falls apart for us is in the accompanying data flags, stereo and surround synthesis, and metadata, where we get stereo and surround hopelessly muddled. Sometimes a stereo signal is flagged as Dolby Pro Logic, so that it will automatically switch to analog surround at the viewer's receiver even if it was not designed for such treatment.

Equally bizarre is the process of synthesizing surround from a stereo signal and transmitting it with surround metadata. This means that when it reaches a receiver driving a stereo system, that receiver will downmix the fake surround signals into the original stereo mix (which may also have been processed, remember), thereby screwing up what may have been a perfectly respectable stereo mix.

This is all entirely hidden from the viewer, who can't tell that anything is wrong technically, and can only assume that TV audio is really not very good (the technical term for this is: "sucks")!

What should happen? As Michael puts it, "The only person qualified to set metadata is the mix engineer for the original project. Although there may be what seem to be good reasons for cheating on this, doing so is not the right way to solve the problem."

Depending on how metadata is reset, it may become impossible to transmit to the end-user what it was that the original producer intended.


We should, as a matter of basic craft, be absolutely meticulous about accurately maintaining metadata and control flags to represent what the signal actually originated as, as opposed to what we would like the end-user to think we're transmitting. We should be equally meticulous about any signal processing we decide to add to any channels of the signal on its way to the viewer.

In the past, it was impossible for those of us in the production suite to control the signal all the way to the viewer's television. With metadata and current production protocols, it is now possible, for the first time. The problem is, we aren't doing it.

In our next installment, we'll take a really hard look at the phenomenon called downmixing, and see what really happens. After that, we'll look at upmixing. Fasten your seatbelts and stuff in those iPod earpods! It's gonna be noisy.

Thanks for listening.

Dave Moulton