I was recently invited to speak to Chantal's "Machinima Mondays" group in Second Life about audio in Machinima. The session lasted for about an hour and I enjoyed it very much. The group, which numbered about 15 people, had good questions and were patient with the inevitable tech issues that come up in Second Life. I chose for my topic, "Working with Voice Recordings", mostly because I was presently working with voice recordings that had been sent to me for a machinima production. 

I wanted to provide a sort of re-cap here on my blog of what I presented to the group. I created a short video demonstrating some of the problems and solutions that come up when working with voice recordings. Let's start with that and I'll add to it afterwards

You can view the video in HD at youtube here

Working with Voice Recordings

• Remember that audio is part of directing as well. Don't wait to work with audio until the end of the production process. Start with audio from the beginning. 

• Learn simple, key concepts of Audio: wav format, digital audio formats, frequency, reverb and noise removal should be the first things you learn. Plenty of info out there on these topics. Try YouTube and the Wikipedia for more info

• Three basic steps to process voice recordings: clean it up, level the volumes and process the tone

• Be sure to level the volumes of the voices. This doesn't mean that every voice has to be at the same volume level all the time, but it does mean that actors will send you recordings that are done at different volume settings and you need to get those levels to sound relatively the same. 

• Clean background noise BEFORE you raise the volume. Encourage your actors to avoid background noise during recordings, but if there noise present the trick to removing it is to sample the noise only then remove just the noise. This is pretty easy to do in Audition and Audacity. Here's a tutorial on noise reduction for Audacity. The problem comes when the noise is at the same frequency as the speaking voice. In this case, when you remove the noise you'll remove some of the speaking voice as well, which you do not want to do. It's much trickier in this situation and would take a much longer post to explain what to do. For now, read up on "Noise Reduction Audacity" via Google and you'll figure it out. 

• Processing the tone of the voice recordings: each actor will most likely record their parts using different microphones with different recording qualities. Your job as the director is to make sure these recordings sound like they were all recorded using the same microphone. The reason for this is that you do not want to distract the viewer of your film by inconsistent or different sounding voice recordings. You want to get the viewer to think that the voices are coming from the visual world you've created in your film. The best way to solve this problem is to use a "parametric equalizer" (tone shaper, basically) and to adjusts the bass, mid-levels and treble of your recordings to get them sounding similar. The technical process is simple in Audacity (check this tutorial), but it takes some time to develop an "ear" to hear different tones. Try to follow the basic rule: voices should be clear and distinct, not too bass sounding and not too tinny. Listen to some of your favorite Anime films or series and you'll hear great sounding voices as examples. 

Finally, it's important that you voice recordings sound good in your film. They share the task of telling the story along with the visual images. If you have poor, or sloppy voice recordings, the story you are telling will be hard to follow as they will distract the viewer. Even if you do just the basics of a lttile noise clean up and volume levelling, you can make your voice recordings sound much, much better. 

I plan on expanding this video to include a voice over and better video. Stay tuned as I hope to have it done soon. 

My thanks to Chantal for inviting me to speak to the Machinima Mondays group.