Ed.note: When I originally wrote this, it was a semi-rant after a long weekend of screw-ups with my automation system, mostly due to producer errors. So the tone is a tad snarky, even if the information is 100% valid. Y'all have been warned...
- Remember that most stations will be automating the playback of your show. Ergo, you must make your show friendly to automation software...and not make your show in such a way that it requires human intervention to broadcast.
- Public radio lives and dies by the clock. Everything is done by coordinated breaks at specific times. If you're not on the clock...and I mean to the exact second...then you've broken format. "Broken" is a good word here because it'll sound like crap as a station's automation cuts off your show because the timing is wrong.
- There is no excuse for not having a fixed length. Hourlong shows should end at 58:00, 58:30 or 59:00. Half-hour shows should end at 27:30, 28:00, 28:30 or 29:00. Pick a length and stick with it! It must remain the exact same length for EVERY episode. If your show isn't be fed live, then you have absolutely no excuse. There's this thing called "Protools" and it's been around a while. It makes editing to time really easy. And it also has time-compression/expansion while is audibly transparent and perfect for getting that last 20 or 30 seconds in/out of your mix. If you are doing your show live, learn to backtime a music bed to fade in under your host's intro/outro. It's not that hard.
- Understand the difference between a audio file's time-length and the program audio time-length. When coming in/out of a break, remember to leave at least a full one second (preferably 1.5 to 2) of absolute silence, centered on the cutaway time. That lets stations "cut away" cleanly without cutting off your program material. This concept is what is meant by a show's "time" being 29:59 (29 minutes, 59 seconds). However, in today's age of ContentDepot and automation, even though there might be 29:59 of program audio (plus the one second of silence) the actual audio file (MP3, MP2, WAV, etc) must time out to exactly 30:00.000 - again, with Protools (or any audio editing software) it's very easy to do. And if you don't, it will screw everything up for the station. If you have four segments and you leave each audio file one second short, that means four seconds of dead air at the end. That sounds bad. Stations don't like it.
- Stations have to pay the bills, that means they need breaks in your show to play underwriting. If you're an hourlong show, at least one sixty second break in the middle is required. Two 60's or two 30's (or some combination) is preferred. Halfhour shows need one 30 or 60 second break...two is nice, but not required. Remember, ALL stations need that last 60 seconds at the end for legal ID's, so that cannot be used for your overall alloted time for breaks.
- Floating breaks suck. Everybody hates "floating" breaks. I don't care how much it "comprises your artistic vision" to have a break right in the middle of something. Deal with it and make your frickin' breaks the exact same time every week. One exception: half-hour shows I don't care as much because odds are good I'm not going to take the break anyways; I'll just let it play through. Actually, one other exception: if you're good about using segments (see point 9 below) then floating breaks aren't as big a hassle. But I still dislike them on general principle.
- Follow the NPR clock for break times. For an hourlong show, the standard NPR break times are 60 seconds starting at 19:00.0 past the hour, and 90 seconds starting at 38:30.0 past the hour. That's assuming no cutaways for newscasts at the top and bottom of the hour. For a half-hour show, there isn't really a standard, but the break should fall around 15 minutes past the hour. If you want to really time it perfectly, call your local NPR station and ask them to give you a copy of the show "clock" of a show you like. Wait, Wait...Don't Tell Me! and Only a Game are good "example clocks" for non-hard-news shows. Weekend Edition and All Things Considered are good for hard-news shows. OnPoint and Talk of the Nation are good for call-in talk shows.
- You don't have to have a cutaway for the TOH newscast, but it's not a bad idea to do it...especially if you're an hourlong show. Some stations will refuse to air a show that doesn't have a cutaway at 01:00.0 and 06:00.0 for the Top-Of-Hour NPR newscast. I'm not one of those stations; but I do prefer to have the option. Still, I admit it's hard to justify it because you know that any content you put between 01:00 and 06:00 might never get aired...so many shows put something of a "throwaway" segment in there (World Cafe plays one music track unrelated to the day's topic, TechNation has the "Five Minutes with Moira Gunn" commentary) but just playing music for five minutes (Living on Earth) can really screw a station that doesn't air NPR newscasts. Having a cutaway for the bottom-of-the-hour newscast is largely unnecessary unless you're a major daily news show meant to air during morning or afternoon drive. Halfhour shows don't need a newscast cutaway so much since it would leave very little time for the rest of your show.
- I mentioned making it "automation friendly" - that means multiple audio files for multiple segments. ContentDepot works this way these days; each segment of the show is a separate MP2 audio file. That way a station can easily program which segments to air and which to ignore (and insert their own local segment for a break). For example, a show like Only a Game has eight audio files: billboard (01:00), newscast (06:00), newscast return music (00:30), Segment A (11:30), Break 1 (01:00), Segment B (19:30), Break 2 (01:30), Segment C (19:00). Add up those times, and you'll notice they total 59 minutes and zero seconds; a perfect hourlong show. What if you are not distributing through ContentDepot, and instead are using, for example, a podcast distributor like Liberated Syndication to distribute your show to stations? In that case, create an "extra" segment that is the assembled show as one audio file. That way stations that want the segments can take 'em, and those that don't care have an assembled show ready and waiting for them.
- Automation is all about "the same"...as in, have your show distributed and ready to air at the same day/time every week. Pick a day and time and stick with it. You can't ever be late, or you'll screw over your affiliates. Have an "evergreen" show (something vaguely generic without any specific, dated, references) ready to distribute if your planned show isn't going to be ready in time.
- Watch your audio levels. This shouldn't have to be on the list at all, but lately there's been too many shows that are produced in someone's basement on a Mac laptop running a portable Protools Mbox and mixed on headphones. This isn't entirely a bad thing; lowering the barrier to entry has resulted in many talented people getting access to the airwaves. Unfortunately, it's also led to a tremendous erosion of quality production values in many shows. One of the biggest transgressions is inconsistent audio levels; where the guest speaks really, really, quietly and then suddenly the host is REALLY, REALLY, LOUD!!! The rub is that there is no "level meter" that really works for loudness. The only way to really make sure your levels are right are to use your ears...and you cannot use headphones for that. Headphones inherently screw up the concept of ambient listening. Plus remember that while dynamic range is great in theory, in reality most FM radios can't handle very much difference between peaks and valleys...and when listening in the car (the place the vast majority of radio listening is done) anything in the valleys is lost in the ambient noise floor. If you lack the audio mastering skill and/or proper studio & speakers to deal with this...then I recommend the "brute force" fix: The Levelator processor. It's a RMS Normalization tool, and actually works pretty decently for people talking. It produces weird results for music, though...and really it's no substitution for a proper studio and proper mixing. Please note, I never advocate dynamic range compression unless you really know what you're doing. Compression gets wonky because every radio station in the country has its own compression scheme going on at their transmitter. And compressing an already-compressed audio source leads to unpredictable sound...remember, using RMS Normalization is "bad" enough - don't make it worse by using compression. The best solution is to use your ears while recording and mixing and make the loudness sound consistent that way.
- Can't spare the bandwidth to send out an uncompressed WAV file to stations? Use MPEG Layer II (aka "MP2") instead of MPEG Layer III (aka "MP3"). MP2 is older and both less efficient at data savings and less flexible than MP3. So why use it? Two reasons: first, it's the standard that the public radio satellite system has standardized on for ContentDepot. Second, it's algorithm is superior to MP3 at surviving cascading algorithms (i.e. later re-compression of the data) which is a real concern with HD Radio which compresses everything down to a max of 96kbps using HDC (a variant of sorts on the AAC codec). There are many free tools out there for encoding into an MP2. I like the PRX encoder, even though it's slower than many of the others, because of its ridiculously-easy user interface and rock-solid results. If you don't use the PRX encoder, remember to turn frame padding off, and set it for 256kbps/stereo or 128kbps/mono and 44.1kHz sampling rate with 16 bits.
- Speaking of PRX.org, it's a good reasonably-priced way to distribute your show if you want public radio stations to be able to find it. ContentDepot is even better but admittedly more expensive. If you're really broke, though, a cheapo podcast from LibSyn or someone else is better than nothing.
- Speaking of filenames, name your segments the exact same filenames every week. The whole point is to drive stations to use the newest episode, not hoard old shows forever and ever. Keeping the filename the same makes it much easier for a station to set up automation; they just tell it to play an audio file named "whatever.mp2" from a given folder (that the podcast aggregator will dump your podcasted audio files into each week) at a specific time.
- Too confusing to have everything with the same filename? If you must delineate the files somehow, put the date you're distributing the show as part of the filename and be VERY VERY consistent in how you do it. For example, let's say you distribute this week's show on May 31st, and you have three segments: first half, break, second half. Your filenames should be: 20080531-yourshow-1sthalf.mp2, 20080531-yourshow-break.mp2, and 20080531-yourshow-2ndhalf.mp2. Important point: the "-yourshow-1sthalf.mp2" part never changes. Only the date part (and I recommend the YYYYMMDD format, it makes sorting easy). That way a station can set up an automatic renaming tool to strip out the date part and copy the new audio file to the appropriate folder for the automation to play it. It is crucial though, that you never change the non-date part of the filename. It must always be the EXACT SAME CHARACTERS (and upper/lower case, too). In that same example, if one week the first segment was "20080524-yourshow-1sthalf.mp2" and the next week was "20080531-yourshow-firsthalf.mp2"? That won't work. Remember, the point is that no human at the affiliate station needs to even look at your show...it just all does it automatically. The only way that works is if you provide their computers with exactly what they expect to see. Otherwise it's "GIGO".
I also recommend Transom.org, too: