After writing about the POS systems last week, I started going through my list of other past lives--things I have done either professionally or as a hobby in the past. This got me thinking about broadcast radio, or more accurately, what Linux could do for broadcast radio.

Although most of my experience has been at the transmitter end--I have a commercial radio-telephone license and have tuned and repaired broadcast transmitters--I am pretty familiar with the other pieces that make a radio station work. Back when I did systems testing on an operating system for big iron (yup, another past life), some of our commercial customers ran programs on our computers to keep track of what music had been played. This was mostly an accounting issue--that is, it kept track of royalties.

Computing has changed a lot since then. Now, computers essentially are free instead of costing millions of dollars. Thus, this simple radio tracking application could be run on any old PC today. But much like the old days of having to write something down and then enter it into the computer, there must be a better way.

The better way would be for a computer to run your station completely. That is, you store copies of all your program materials on the computer, and then tell it what to do when. Materials can include music, pre-recorded programming, commercials and practically anything else. Toss in a few voice features, such as a digital voice announcing the current time and the output of a weather station, and the only time you need a human is when you have a live program.

Expanding a little on the "what to do" idea, there really is no reason you need to be too specific. You could, for example, tell the system to select randomly from a pool of songs, never playing one more often than every 27.5 hours.

This radio design isn't far-fetched; in fact systems like this already exist. For example, Radio Paradise pretty much has this radio model in place--with Linux. In fact, Radio Paradise is well worth a look and listen simply to see how such a solution works. But, a lot of radio stations are out there, and Radio Paradise is the only one I know of that's running the automation on Linux.

So, what's wrong with this plan? I have two local (that is, Costa Rica) examples to draw from, and both are shortwave stations. There is nothing special about that fact; shortwave stations simply happen to be the stations with which I have talked. In reality, an Internet-only station could use the same software for its work.

Back to the examples. First, right up the street from my house is what is supposed to be the world headquarters for Adventist World Radio. I put a slight disclaimer here because the sign on the street announces this, but I didn't discuss it with anyone at the station. In fact, I was at the station for an auction of radio equipment when the following discussion developed.

The person I talked to, one of the few that spoke any English, said the station was using a station automation package that runs on Microsoft Windows. He then went on to explain that the system wasn't reliable enough. That is, it would fail and someone would have to reboot a computer--which pretty much defeats the purpose of station automation.

Of course, I bought up the idea of using Linux. To my surprise, he was familiar with Linux and said the station had tried loading it on one of its computers. His big concern, however, was that if the station found a Linux-based solution it would have no support. That is, what if Linux failed? Or a possibly free software package? In his case, he was a one-year volunteer; even if he got up to speed on the solution, he would be gone within a year.

The second example is Radio for Peace International. RFPI is located about 20km from where I live, but the back-haul to its Internet connectivity is supplied by a wireless link whose antenna is about 20 feet from me as I write this. I am much more familiar with RFPI's operation and expect it is typical of many stations.

RFPI downloads much of its program material from the Internet--using a Linux system, of course. Rather than save the material on the computer, RFPI saves it on mini-disks. Broadcasts, then, are done with a live announcer filling in between pre-recorded material. The live broadcast also is recorded on tape for re-broadcast later in the day.

In the RFPI example, the only missing link to full automation is some software. That is, the program material already is in a form that could be saved on the computer, and that same computer certainly is capable of doing the editing to add the local content.

Why doesn't RFPI make this change? Simple--it already has something that works. Making this change to full automation would require procedural changes and training. And, when the automation breaks, the station needs someone to bail it out.

I think the answer is yes. The technology to make this change to full automation already is available. You can see examples in place today, such as Radio Paradise. I also expect, but can't be certain, that some very fancy and very expensive commercial products are out there, being used by the big stations.

Why would someone use one of these expensive commercial products? Because they work. I think the answer really is that simple. The cost associated with a commercial radio station being off the air is lost advertising revenue. Being off the air for a few minutes could equal the cost of an entire automation system.

Is there a good market here for open-source software? I think so. I see two products, the software itself and the support. An open-source solution used by stations around the world would have the advantage of more testing and, thus, higher reliability. The important part for radio stations, however, is support. If the systems go down, they need to be fixed and fixed fast. This is where the revenue side comes in. On the low-end, this may mean that someone from a computer store comes out and fix the system if it breaks. On the high-end, this could mean in-house staff on site 24/7 to make sure nothing breaks. You can pick your business model anywhere in between.

I would like to encourage comments. Although I don't have much free time to work on this, it is a topic in which I have a lot of interest. With a little bit of organization, I believe there is a lot of potential for Linux to take over radio station automation--one more step toward world domination.

Phil Hughes is the Publisher of Linux Journal.