As you might know by now, I started an informal podcast in Croatian about Free and Open Source software called Razgovori o slobodnom softveru. This blog post will try to summarize my experiences and serve as information for podcast guests on how to get the best possible audio quality.
Every podcast currently starts as google wave. While this isn't ideal, it does work for now. I would love some other service which would allow me to see information color-coded by users. Those notes are later used to write blog post with all links and information about episode.
First, let's get Skype out of way. Only reason why we aren't using free software for audio conversation is simple fact that everybody has Skype and there is great skype-call-recorder which allows us to get uncompressed audio recording with local audio on one channel and remove one on another which enables easy matching of audio levels for local and remote speakers.
While we are at audio levels, let me share a little secret: right after recording audio from Skype, I use Levelator to adjust audio levels. There is Linux version of it, but it doesn't seem to work for me on recent Debian. But, effect is so valuable to good quality podcast that I just use Windows version under wine.
As far as guest audio is concerned I have found following notes really important for my guests:
- use any microphone other than one on your laptop. Built in microphones catch typing sounds and all huss and buzz from machine itself. Even cheapest possible microphone will generate better results, just compare Ivan's audio from episode 0 (audio via built-in laptop microphone) versus episode 1 (which is cheap plastic external microphone)
- if you have headset with microphone, use it. Adjust your microphone level to be above your nose. That will prevent recording of air that you breath.
- When you need to take a pause while speaking, just do it! I edit whole podcast to remove bloopers and breading anyway, but it helps if you repeat last word of full sentence when you start over so it will fit in context (and intonation) of previous material.
Much more useful information can be found in Skype for interviews by Doug Kaye and Paul Figgiani (guys behind Conversations Network which also implemented Levelator mentioned above), so if you need to record on Windows or Mac take a look there.
For editing audio I use audacity which is great audio editor, but I still have to learn all keyboard shortcuts because using mouse to edit audio is too slow and cumbersome if you want real flow of montage (and I did video montage back in 1990 on local TV station, so I know unnecessary mouse clicks when I see them). It would be nice if labels moved when I do cut of audio, because editing of audio leaves labels at old positions which isn't useful since audio moved. To give you some idea of how much effort is involved in editing audio (mostly removing silences, bloopers and breading): for last 35 minute podcast, it took 420 edits to get it into final state. And most of single day. It's about the same as creating good presentation: I often need 6 times more than presentation length to create it. For audio editing, you can count on 10-15 times ratio for editing.
Finally, I hop over to some creative commons audio site (to be honest, it was first one in google search for CC music), pick one of top tracks and extract intro and outro music from it to put it at beginning of podcast and end. Finally, audio introduction and web address information for end are recorded and put on top of music.