Creating professional recordings for IVR phone systems, using Audacity

Most, if not all of you will have encountered an IVR (Interactive Voice Response) phone system at some point in your lives.

“Hi, and welcome to Larry’s Cheese Emporium. To order cheese, press one. To cut the cheese, press two…”.

Ahem, yes…

Well today we’re going to show you how to create your own professional-sounding IVR recordings. All you’ll need is a microphone, a quiet area, and a copy of Audacity.

NOTE: This tutorial assumes a basic working knowledge of the Audacity Digital Audio Editor (audacity.sourceforge.net). If you’re a beginner, there are plenty of tutorials around on the net that will get you up to speed in no time.

Before we begin, there are a few points to be noted:

• Always take your recording in the highest quality your hardware can manage; this leaves us with a good quality foundation for our editing.
• We’re going to resample our recording down to 8kHz. Most phone systems still only support 8kHz sound; anything above this is sampled down. So if you upload it at 8kHz already, you reduce load on the server, as well as ensuring that it sounds the way you want.
• Do your initial recording at 48kHz if your hardware supports it. 48kHz is evenly divisible into 8kHz, which means the conversion to your final output is cleaner. This helps keep the quality high through the conversion process.

Start by creating your recording in Audacity:

First of all, you will want to make your recording in full quality by default. As was said above, 48kHz is preferable if your audio card can support it.

Start by recording about 4 seconds of silence. This will be used as a baseline for noise reduction, which will help to remove any unwanted noise.

Next Run Noise Reduction -> Get Noise Profile to learn what ‘silence’ sounds like in your setting (it’s NEVER completely silent).

Delete your track of empty noise, so that you are left with a nice platform to record on. Record your track.

After recording, always run Noise Reduction before running anything else.

Add a very subtle echo to fill out the sound. I used a delay of 0.01, and a 0.1 decay rate

Normalize the sound now to avoid clipping. Clipping is when audio becomes distorted because the amplitude of a track is too high for an amplifier to output, causing the audio wave to flatten out or “clip”.

Run the equalizer. For phones, I raised the low to mid-low frequency, and chopped the high frequencies down. This just made the sound a bit less harsh and grating when heard over the phone.

Run the normaliser again to get it to a good volume that won’t blow people’s ears out. We wouldn’t want to do that, now would we?

Take the time now listen to the recording and trim out the unwanted bits. For example, the silence at the start & end of the track, sounds of swallowing, breathing, hesitations, etc. This may take some time, but the effort will result in a much more professional recording.

Listen again to make sure the track doesn’t sound unnaturally clipped due to the previous step

Finally, save your project. This is “full quality”, and is as good as you’ll get.

To create the sound file for the phone system:

Set your “Project” sample rate (at the bottom) to 8kHz.

Use the Resample function to resample your track to 8kHz.

Convert your track to 16bit PCM bit-depth.

Click File -> Export, and export the file as “WAV (Microsoft) Signed 16bit PCM” file format.

And there you have it! The audio track is now ready for use in your IVR phone system.