Cleaning audio

SECTION ONE: PROCESS OVERVIEW

TITLE: What are you calling this process?

> Cleaning audio

SUMMARY: A one-sentence description of the process.

> Cleaning the audio to have better quality sound.

WHY: Briefly explain why the process is important.

> We want to go through a process of cleaning the sound in the video so the voice is clear and the content understood.

INTERVAL: How often does this process need to be complete?

> Whenever a new record audio is available for cleaning.

SECTION TWO: PROCESS EXECUTION

STEP #1

Split stereo tracks into Mono

Sometimes audio will be recorded in stereo when it is mono, like this:

In this case, split the tracks into Mono and delete one:

STEP #2

Reduce background noise

Find a section of audio where there is no talking so it’s just background noise:

Select the audio and using the Noise Reduction (Effect - Noise Reduction), get the noise profile, then select the entire channel and reduce the noise over the entire audio recording in that channel:

Do this for each channel (get individual noise profiles because some channels will have different levels of noise).

If there is still noise, may need to run this process multiple times.

Also look for sections of audio where the noise profile changes, for example, a hum when a fan/air-conditioner turns on:

STEP #3

Compress audio

If the level of audio is quite low, like this:

Compress the audio to increase the noise levels:

I use these settings in the compressor (Effect - Compressor):

Note: After each round of compression, check the noise level again, you may need to do another noise reduction step if the background noise becomes audible

STEP #4

Careful when using compression

Don't be too heavy on the compressor, it can making the voices sounds a bit funny, plus doesn't have to squash it so much

STEP #5

Remove unwanted noise (coughing, breathing etc)

Sometimes you will hear coughing, typing etc in the background of one audio channel, it is often visible like these sections between the speaking parts:

There are a few ways to do this, I usually just select the section of audio and reduce the volume on it, but make sure that the audio isn’t actually required, like someone saying ‘yes’ or agreeing with the other person should remain.

STEP #6

Other things to look out for:

• Pops


• Clicks


• Low frequency rumbling (Can EQ it out, anything below 100Hz should be ok)


• Heavy breathing (especially if audio has been compressed)


• Sneezing/coughing/clearing throat etc


• Extended pauses


• Too many “Umms…”


• Watch out for bleeding of one audio track into another (usually coming from the headphones of the other party)


• Flow of the conversation – sometimes the speed of conversation can be slower while people think of what they’re going to say, particularly if English is not their 1st language, so we can speed up the flow of conversation slightly and make it more fluent by making small reductions in the gaps between words and phrases, but be careful that it still sounds natural after editing.

STEP #7

Remove any low frequencies below 60Hz

It won't really change the sound of the voices much but when we get a pop or some surge of air in the microphone, it wont sound so bad because the low frequency thud will be removed

It's very noticeable with ear buds.

SECTION THREE: REPORTING

WHO: Who should be the outcome sent to?

>

WHAT: What are you expecting to see in the outcome?

> A clean audio file

WHEN: When does the person reviewing the outcome want it?

> As soon as the raw audio file is available

HOW: How should the outcome be delivered?

> Audio file edited