I’ve done quite a bit of basic audio production: recorded talks, group discussions, video background music and so on. I know enough elementary facts about high and low pass filters, sample rates and so on to get through a related conversation. But is that it? One thing that baffles me is that some people going way over the top with the file size of their output, as if that will improve the audio quality.
So — what file size should my podcast be?
Well, look at 5by5.tv — the network outputs dozens of shows a week and the audio quality of their talk shows is brilliant. That’s primarily a byproduct of “production values” in the studio, but you can’t turn lead into gold (even with gold plated leads), your crummy recording off the church sound system’s auxiliary ports won’t sound better by giving more kilobytes to the final mp3.
Dan and Haddie have a great setup, noise isolation loops, compressors, limiters, etc. Their production quality comes from the gear they’re using (and engineering knowledge), but look at the bit rate and file size. This is a 47 minute track, and is just under half a megabyte per second. That’s half of the very minimum you would ever use for music.
For what it’s worth, This American Life — the quintessential podcast — uses the same settings.
So, 64kbps downmixed mono audio is more than enough bandwidth for your spoken word. Don’t use stereo unless you’re producing an audio play that heavily relies on it, but even then, consider going mono. Anything more is a waste of bandwidth for you, your mobile listeners, and of disk space for your regular listeners.
If you’re making a music podcast, you’re probably either: a) a professional radio producer, or b) a bedroom wannabe-dj. Which will affect both the quality of your source (legal CDs/high quality lossless audio/high quality MP3s, and standard quality MP3s, respectively) and your production levels. iTunes MP3s come in at 256kb/s Stereo (~2Mb/minute), whenever you re-compress that audio you’re going to lose further quality regardless of method, so sticking with 256kb/s may be futile in order to save audio fidelity.
In my experience music based podcasts are meant as mini-showcases of music, intended for review and promotion, and so sticking to a reasonably efficient bit-rate would be preferable. 128kb/s or 172kb/s should suffice. Stereo isn’t necessary, mono will allow you to pack more punch per kilobyte, but if you’re straying to the higher end of the byte range, make it stereo.
This one’s very much to the publisher’s discretion, but I would say stick to 128kb/s. Bandwidth is precious, and if people want the highest (FLAC, 320kbps+) quality music they’re not going to be satisfied with what you can squeeze into a downloadable mp3 anyway.
For example, All Songs Considered’s “Tiny Desk Concert” series outputs at 128kb/s “joint stereo.”
- The BBC have published lots of technical requirements here.
- Richard Farrar posted an article with a similar intention a couple of years ago.
Any comments or suggestions? I’m definitely not an expert, as I said above, just know how things have worked best in my experience. Would love to hear your feedback.