Description

During my volunteer work at the local TV station I noticed that a lot of editors were having trouble recording a voice-over for their news item. Some of the common problems were:

• Improper gainstaging (where the audio is either too loud or too quiet).
• Unevenly balanced timbre (with either too much or too little gain on certain frequencies)
• A large dynamic range (where the difference between loud and quiet signals is too great for broadcasting purposes)
• Two people necessary for recording a voice-over (due to the recording room being separate from the editing room)

These problems could either arise due to incorrect microphone placement and/or incorrect use of hardware/software. My aim was to create a tool that would solve all of these problems.

Recording process

To overcome the problem that required two people to be present to record a voice-over, the entire application had to be controllable from within the recording booth. We put a pc in the edit room with cables running to the recording room so that the room would stay quiet. I also connected midi controllable switches for the recording indicator outside the recording room and repurposed an old keyboard to be able to start the video on the edit computer. This way, the video would automatically start playing whenever you'd press record in Voicesoft to speak along with the image.

Audio Engine

Aside from making the recording process as easy as possible, providing a solution to the issues with inconsistent audio as good as possible was most important. I created an audio engine in a separate window that could only be accessed trough a certain command to provide a set of audio processing tools you'd typically find on each channel of a digital audio mixing console. An eq (or equalizer), a compressor, noise gate and a limiter. Prior to recording their first voice-over, I'd do a brief calibration of each person's voice and save the settings in the audio engine. People could then select their name in the main window and the specific settings for their voice would automatically be loaded in. This solved the issue of having to calibrate the sound each time a voice-over had to be recorded.

Inner workings

Because I completely built this project in Max/MSP (similar to Pure Data), it usually isn't called "real" programming, but the great thing about these visual programming tools is that you can prototype really fast, so you find out really quickly what works and what doesn't. This is one of the many projects I created in Max/MSP and it allowed me to create whatever I could think of really fast, and use that as a basis for porting to a more sophisticated language, like C/C++.