Audiobook Recorder

When I started narrating audiobooks I took the normal path, that everyone takes, of using Audacity. Now, don't get me wrong, Audacity is a fantastic program and a joy to use, but when it comes to recording and editing audiobooks it really started to get me down. I was just making too many mistakes and having to go back through every chapter and cut out all the outtakes, and find all the really bad lip smacks and breaths and remove them took longer than the actual recording of the audio.

So I thought to myself "there has to be a better way of doing this!", and of course there was.

I had the idea of recording each individual sentence as an individual file and writing a program that would take those files and work out where the start and end of the speech actually was. And then I thought it would be good to be able to re-record the last phrase instantly without having to do any editing. So I wrote a program, originally for the Raspberry Pi written in C and SDL (to run on a small 2" screen).

And it worked well. But it was awkward to use.

So then I decided to migrate the concept to Java and run it on my PC (I was originally wanting to use the Pi because it was silent and fanless). And so that's what I did. It allowed me to create a much better GUI for it and also make it cross-platform.

Since then the project has taken on a life of its own. So many useful things have been added that it now pretty much does everything for you. All you need to do is provide the voice.

Some of the cool things it does:

• Record phrases at the touch of a button with pre-defined inter-phrase gaps (sentence, continuation, paragraph and section)
• Perform either FFT or Peak detection to find the start and end of the phrase
• Automatically trim the audio on a positive zero crossing for click avoidance
• Insertion of "room noise" randomly selected from a 5 second snippet
• Stitching of the phrases together into chapters and export as MP3
• Automatic Speech-to-Text of the audio for rapid navigation of the recorded phrases
• Application of various effects (biquad filters, delay line, AGC, clipping, amplification, etc)

And the one underlying ethos of all this is "be non-destructive". That means that nothing ever gets done to the audio files themselves. Those always remain the exact files that you recorded (with the exception of using an external editor or effects processor, or cutting and splitting of files). All effects and trimming are applied realtime when playing or exporting. That means that if you change your mind about an effect you want to have on a snippet, or you want to change the gain of it, etc, you never lose any quality. Nothing is final.