Buy “Podcasting Church” here. If you’ve read it as I wrote it, why not buy a copy for your pastor or tech leaders? If you just want to be nice, why not spend the $ to help me out. If you hate me, why not buy 10-15 copies to burn? That would REALLY show me! Whatever your reason, buy one now.
I’ve had an idea for a book in my head for a few years now. Since there is was a challenge afoot, I started, November 2, 2010, to write that book. This blog will contain the very rough draft and I hope I’ll get a chunk of it out this month. No guarantee I’ll write everyday, but I hope this won’t be the last entry.
Them Bits Hertz
If you’re not a detail person, if you don’t care about the geeky pieces, this might not be the chapter for you. I just couldn’t get past the need to talk about bit rate and sample rate.
When it comes to audio encoding, these are really the two most important pieces that you need to know. If you’re paying for your bandwidth, you’ll want to encode at adequate quality, but not so much that you’re wasting money.
Also consider your audience. Larger files take longer to download and more space to store. This is much less of an issue now than when I started podcasting in 2005. Even so, some of your audience have older computers, slower internet connections and smaller mp3 players. Consider them, too.
Sample rate
The way that digital recordings are made are is different from analog recordings. An analog recording was an analogy of the sound (or video for that matter) recorded to some medium. A digital recording takes samples which are then reproduced to recreate the original audio. More samples creates a more faithful reproduction to the original.
This rate can range from very low (8 kilohertz) to better than cd quality (48khz). The difference between 8 and 48 might not seem like that much until you realize what’s being measured. A hertz is a unit of frequency. In the U.S., our AC electricity cycles at 60 cycles a second or 60 hertz. For mp3 encoding, we’re measuring in thousands of hertz. A poor recording might still have 8 thousand samples a second while an amazing one would have 48,000. CD’s are recorded at 44,100 khz by way of comparison.
Bitrate
This is only half the equation, though. It tells us how many samples are being used, but not what size those samples are. Think of it this way, if I take a picture of a sporting event 48,000 times a second, I would have a lot of information. It wouldn’t be all that useful, though if the resolution of the picture was 200×300 pixels (much less than an average television).
If I took the same number of pictures as a much higher resolution (say 1080p or even 4k like some movies), I’d have much more information that’s much truer to the original, even providing more information that a human can process (most movies and video are captured at 24 to 60 frames a second, not hundreds or even thousands).
Bitrate answers the question of how large the sample is. An 8 khz, 320 kbps audio file is no better than a 48 khz, 16 kbps recording. The first takes too few samples. The second, takes samples that are too small.
People make entire careers off of compression, so I won’t try and sum it up completely in a short chapter. I will say that presets are a good guide of where to start. Tweaking bit and sample rate just right can result in quite small files that are quite listenable.
This also comes down to content and audience. A recording of an orchestra and choir requires much more data and skill to reproduce than a single person talking. Where your audience listens is important too. A home listening room with expensive speakers will produce a much more faithful reproduction than an mp3 player hooked into an old car’s cassette deck connected to two blown speakers.
I always released a mono mix of my podcast at 22.050 khz 96kbps. When you’re just one guy doing spoken word, you don’t need any more. Some would even argue that I was recording too high a quality show at that rate.
VBR or CBR
All this discussion has assumed a constant bit rate, but that’s not really necessary. A 1/2 second of silence doesn’t need as much data to reproduce as other things. Wouldn’t it be great if you could tweak the settings on the fly to create a better recording for more detailed sounds and a lower quality recording for much simpler sounds?
That’s what a Variable Bit Recording does. It intelligently tweaks the recording to create a smaller file size by recording simpler sounds with less detail.
Returning to our video example (although it’s not exactly correct), if you were recording something with moments of black or solid colors and moments with tons of detail, you’d want to dynamically change how much information was recorded based on the amount of information in the original. VBR does just that.
To achieve this, there are two methods. The faster of the two is “single-pass” encoding. The slower, but better method is “multi-pass” recording. In a single-pass recording, the software takes a crack at removing unnecessary detail once. Multi-pass recording as the name suggests, takes more than one pass at the recording to create an even better VBR recording.
This all comes at a cost that you should consider based on your time, source material, equipment and audience. The smallest file will be a VBR, multi-pass recording. It will also take the most time to create. A CBR recording will be larger, but take less time to create. In the middle is a VBR, single-pass recording. Experiment and see which fits your needs.