Audio Diaries – Tough Talk

Some time mid last year my friend Sam got in touch asking for help with some audio files for the first episode of his upcoming second season of Tough Talk, a web series he’d started devoted to mens mental health here in New Zealand. As his work is something I believe in the value of, I elected to help him out, and had him send over the files.

The raw audio I received was in a pretty poor state, and far from fit for purpose. The voices were extremely quiet, overwhelmed by background noises of all varieties, and I wasn’t sure I could salvage them. Still, they were all Sam had to work with, and couldn’t be re-recorded, and they gave me an excuse to develop my audio restoration skills, as well as get some practise in post production. By some miracle I managed to get the tracks to a fairly useable state, Sam was pleased and relieved, and we made arrangements for me to work on the remainder of the series.


Sam & Kristina setting up on set

Over the course of the series, I found the files being sent my way to be in varying degrees of quality. Most of the interviews had been recorded by the time I was first contacted, but some of them yet remained, and with some general advice and pointers, Sam and his partner Kristina produced higher quality results. By the end of the seasons filming they’d learned a lot about how best to work with sound in different, and challenging environments, and I’d figured out a thing or two about how best to get useable results out of the audio captured under those less-than-ideal circumstances.

The first major lesson I learned in restoration is that blunt-force processing won’t get you anywhere. The biggest problem I had was dealing with background noises of all varieties, ranging from the white noise of the lapping waves, to the brutal humming of industrial electronics and machinery, and a variety of sources in between. Heavy settings on de-noising processors would result in lifeless recordings, and many of the subtleties of the recorded voices would quickly be lost, taking with them their intelligibility.

I solved this by tweaking the de-noising processors to have very gentle effects, so much so that to the untrained ear you probably wouldn’t notice anything had happened. I’d run the audio through these subtle settings, then re-tweak them, and run the processing algorithms again. Depending on the severity of the background noise, I’d do this anywhere between 2-4 times, always aiming to stop shy of any loss in the desired parts of the signal, the voices themselves. The software I used (iZtope’s Rx) had a useful feature wherein it intelligently scanned a selected part of a file and developed a spectral profile of the noise, which helped it identify what it was tasked with filtering out. This helped, but I noticed that most critical was carefully setting the threshold of the background noise, and avoiding any overuse and overkill. The end results weren’t always perfect, but they were a notable enough improvement, and made EQ/compression settings easier later on.

Removing static humming was more straightforward, and another specialised algorithm would detect the harmonic series of the unwanted part of a signal, and allowed it to be filtered out via familiar threshold settings, much like a standard noise gate. Generally, these processes were all that were necessary in getting the files ready for mixing. It has always been my first port of call to remove the parts of an audio file that one doesn’t need, before one proceeds with any sort of standard processes. Even before these algorithms were applied, the first things I always do is filter any unnecessary low end frequency content with the cleanest high-pass filters in my arsenal.

Once all this was done, I’d typically go ahead with standard mixing techniques to provide Sam with a stereo file to match up with his footage. I tried to avoid over-compression wherever possible, so as to avoid raising the noise floor again, and relied instead on FabFilter’s Pro-Q and Pro-DS to further sculpt the audio, before end-stage panning and levelling. Simple processes, but when dealing with sketchy source material, I found it’s wisest to avoid the temptation to throw everything at it. You’re only going to manage to take things so far, so don’t try and stress yourself over achieving impossible results.

Some time in the future I may write up more of a tutorial on these processes, but for now, these were the findings I took from these efforts, and you can hear the results for yourself here. Season 3 is said to exist somewhere on the horizon, so until then, enjoy.

This entry was posted in Uncategorized. Bookmark the permalink.

1 Response to Audio Diaries – Tough Talk

  1. Gareth says:

    This is a really awesome read. And an interesting insight into audio recovery. Makes sense to do reductions in small increments. Similar to an unrelated topic, but small increments of distortion for certain purposes.

    Thanks for your post. I’ll listen to some of the podcasts Sam recorded. Mens health is an important topic, I struggle with it too.

Leave a Reply to Gareth Cancel reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s