The “Break it Down” Blog

This write up should help you get together what you need in order to put together and record a podcast that includes folks connecting over Skype.

The information from this blog was the result of me digging across the web trying to find a good way to record and create a podcast that included me talking on my headset to people I was interviewing over Skype. I know the solution I wanted to find would be simple and seem intuitive. The solution I did not want to find was the one that had me installing device drivers and editing system startup files on my computer.

On my initial pass at this problem resulted in a lot of tutorials talking about installing Soundflower, which as far as I can tell is a virtual device driver that allows inter-application-sound-redirection. I think that means it’s a virtual device that allows you take one application, like iTunes, and have it “play” all it’s audio out to the Soundflower device, then you can setup another program to record from the output of the Soundflower device. That sounded OK to me, but not great. Why am I installing fake devices? Why the hell can’t I just hit record and be done with it? Well maybe I can…

After trying to get Soundflower working and giving up I did notice that all the tutorials consistently mentioned the Audio Hijack Pro (AHP) application from Rogue Amoeba. So I went ahead and bought this fantastic piece of software. It’s very professional and simple to use with plenty of fine tuning controls and functionality to keep anyone happy. After playing with AHP and Soundflower for a while and not getting any of it either working or working well I decided to step back from the problem. This has got to be a common problem people have, I can’t be the first. So I went back to Google and started searching. I ran across a seemingly simple utility called Call Recorder which is a little call recording utility that plugs into Skype. It was only $12 so I bought it. Within 5 mins of installing it and trying it I erased it, knowing I hadn’t found the solution yet. The quality of the recording was junk, the program consists of about 3 options and the Quicktime .MOV to .MP3 converted crashes every time I try and use it. Looks like it’s back to the drawing board again.

At this point I was really scratching my head. Of all the functionality in AHP I was certain I could do what I wanted with AHP alone… why did I need Soundflower or any other fake device driver? I just had a gut feeling about this. So I opened AHP back up again and set out to make this damn thing work. Within 6 seconds I realized that AHP has not only recorders you can setup to listen to devices but recorders you can setup to record from applications. I added one for Skype with about 2 clicks of my mouse, then set it up to crap out the audio stream to a specific directory:

I then setup another recording for my headset to also go to the same directory:

From there I tested the setup by calling my wife and recording both tracks: my headset and Skype. The audio quality was fantastic and I had both streams individually which made it easier editing it all in GarageBand because I was able to amp up or subdue either of the channels depending on how each sounded.

When I originally tried this setup I just had it record my headset and then added an “effect” of also blending in the output from Skype (I told you AHP is cool) into a single file:

The problem with that was the person with the headset (me) sounded about 10x clearer and about 5x louder than the person on Skype on a land line (my wife). So I decided to split the tracks into two files so in GarageBand I would be able to tweak the volume to have everything match a little better. While recording from AHP I also decided to tweak the recording quality settings to max them out, so my “source” files were of the highest quality. I could always downsample them in GarageBand or any other program if the resulting podcast was too big.

When the interview began I went ahead and clicked Record on both of my sources. From what I can tell you can’t instruct two sources to start recording at exactly the same moment, so it’s possible that you might have one track a few ticks ahead of another one. This just means when you drop the tracks into GarageBand, you can’t start them both exactly at time 0. You might have to scoot the track that started recording a moment later over to the right a little as it won’t have as much lead time on that track:

The few times I’ve had to correct for this it was about 30 seconds of work. Just find a part of the podcast where you are talking freely with the other people and scoot the shorter track around until it lines up the responses correctly with the people over Skype, not hard at all.

After the interview is over I clicked stop on both tracks. I then popped open GarageBand. The first thing I did was to remove the Female track and the Radio Jingles track (I didn’t need them). I then duplicated the Male voice track and dropped both audio streams (headset and Skype) onto the two tracks (we both happen to be male and I wasn’t sure if GarageBand applied any pre-filters to “male” or “female” tracks so I just duplicated the male track to be on the safe side). Then I dragged and dropped a jingle from the default audio group onto the Jingle track to give some nice background music to the podcast:

Something I noticed right away, even though the jingle track was set to “Duck” whenever audio comes up on other tracks, is that the music was a bit too loud and I wanted it to fade out after the introduction of the podcast. I didn’t think it would sound good if the music was playing on repeat the entire time in the background. I wasn’t sure how to do this but after a sufficient amount of clicking around and checking the help I learned that you can fully manipulate the volume of any track down to the fractions of a second with a really nice click-drag-control-point type of schema:

After fixing how I wanted the music to fade out at the beginning then fade in at the end I decided it was time to finalize this whole thing. First I had to set the max length of the podcast, which is a bit strange. Decide where you want the podcast to stop, in my case it was exactly at 20 mins. However since I had to drag another copy of the jingle onto the “Jingle” track towards the end of the podcast (total play time of the jingle was about 2mins), that means that I had about an extra minute and a half of the jingle hanging past the 20min mark I wanted to end the podcast at. Initially I thought it was good enough to go to the end of the GarageBand timer, grab the purple slider and drag it down and let it chop off everything at 20 mins. Much to my frustration this doesn’t seem to be how it operates, instead it kept popping back to the end of the Jingle track which was about a minute and a half longer than I wanted the podcast to be. So I went to the end of the melody, when I got the resize mouse cursor I clicked and shortened it to exactly 20 mins, I was then able to drag the purple end marker over to end the podcast at 20 mins:

Next I exported the podcast which creates an M4A file. This works fine with iTunes but for folks that listen to “podcasts” by clicking on MP3 files, I tried to find an easy way to convert the M4A to a MP3 file. I googled for “GarageBand convert m4a to mp3″ and was surprised to find a lot of useless links. I think this is a generally good rule of thumb… whenever I search for something, say a problem I’m having, and I can’t find any pertinent links to it right away or very very few that usually means I’ve somehow turned a non-problem into a problem. This is usually caused by missing some obvious setting somewhere or some feature that does exactly what I want. In this particular case the solution was a little more vague.

I found plenty of commercial 3rd party apps that would convert my M4A to a MP3, but I refused to accept the fact that on a Mac I cannot convert a M4A to a MP3, so I kept digging. After a lot of forum hunting I came across a way to convert the files to many different formats in a less than stellar way, but it worked for my purposes. Open up iTunes, go to your Preferences then to Advanced then to your Importing tab. From there set your Import Using to which ever format you want to convert your M4A file. Additionally you might want to set a higher quality setting than the default. Now go ahead and play your M4A file from iTunes so it’s imported into your library. Then be sure to right click on the file and select “Convert to”, in this case “Convert to MP3″. After that operation is done right click on the file and select Show Song File. From there you should be able to copy this file to any directory you want. While I agree these steps aren’t very smooth and it would be about 100x nicer if GarageBand could just export the file to the format specified in the first place it did provide a solution for me without buying any extra software.

Well it looks like we have gotten to the end of the this tutorial. To recap this is all the equipment and software I used to product my podcasts:

• MacBook Pro
• Plantronics Headset with DSP ($50)
• Skype
• Audio Hijack Pro ($32)
• GarageBand
• iTunes

You’ll notice that if you have a decent headset or microphone the only portion of this production that cost money was the Audio Hijack Pro license. I hope this tutorial was helpful, I look forward to any feedback you all have. (Digg this)

UPDATE #1: I just found an extremely frustrating bug in GarageBand when trying to import audio for tracks that have long file names that are very similar to an already-imported track (more than 24 characters or so). GarageBand will simply duplicate the track you already imported. For example, I had some podcasts where I stopped recording and had to start recording again. So for a single podcast that meant I had two audio files that represented my audio stream. Since my device in AHP was named “Plantronics Headset”, the file ended up being something like Plantronics Headset - 20060512 - 1123.mp3, the problem is I had another file almost identical with the last 4 numbers different. When I would drag and drop the 2nd track into GarageBand, it would just duplicate the first track. I eventually renamed the files to just Headset - 20060512 - XXXX.mp3 and removed the Plantronics portion of the name, which fixed the issue. Just be aware of this incase you run into this problem, it’s maddening.