Basic Cleaning up and/or Sweetening Audio
I’m trying to put something together for my students that will help them with basic sound clean-up. What I’ve found has either been not comprehensive enough or too technical. I’ve cobble the following together from several sources but because it’s not my area I’m not sure that its all correct, complete or in practice irrelevant.
Can someone please have a look and either make comment, or direct me to a resource that can help me.
Sorry its such a long post but if someone can help me to get it correct I'm happy to post it as a tutorial. I think there is a need. I certainly would have found it useful
Basic Cleaning up and/or Sweetening Audio
The best way to end up with good audio is to get a good quality recording in the first place. However conditions may not allow for this and the resulting problems need to be fixed, and even the best recording may need to be ‘tweaked’ to give it the exact character you want.
There are many ways to do this. What follows is a summary of the most important.
1. Normalize the signal
2. Remove noise
3. Cut off high frequencies
4. Compress the dynamic range
1. Adjusting Volume and Normalization
• Normalizing: All sounds are boosted equally till the loudest reach the maximum possible without distortion, ± 0dB. If its possible to set this manually it’s suggested you set it to -2 or -3 db as this will not make a perceived difference to the max loudness but will lesson the chance of problems.
• BTW: Most sound in a file should be in the -8 to -14 ranges with the peaks getting near +-0 db, but this is only a guide. You should also consider the nature of the sound itself, e.g. some things are naturally quiet. You also have to consider the volumes of adjacent shots. Too large a difference and it will distract the viewer’s attention.
2. Removing Noise
• Removing occasional noise
o (Sound booth) Transient noises show up more clearly on the waveform. Use the time-selection tool to select
o Noises, like the cell phone ring easy to spot and eliminate in the spectral view. Use the marquee too to select
o Once selected click use ‘Auto Heal’ to remove
Removing Consistent Noise and Hums (Sound Booth)
• Soundbooth’s Noise filter.
o Click the ‘Clean Up Audio’ group
o Highlight a short region that contains only the background noise and click ‘Capture Noise Print.’
o Click ‘Noise’ make sure that the Use Captured Noise Print checkbox is selected
o Adjust the ‘Reduction’ and ‘Reduce By’ Sliders to produce the desired effect
o Use Preview to fine-tune your settings, using the green button to the left to toggle the filter on and off.
o Remove as much of the noise as possible without introducing distortion
3. Remove High Frequencies
• The human ear can hear sounds up to the 20kHz range
• Sounds in that range tends to be very harsh, so apply a cut-off at around 16kHz.
• If the scene has a particularly high-pitched sound like a whistle, or bike bell, you may push up as far as 17.5kHz to maintain fidelity with those particular sounds, but no higher.
• A cut-off filter is applied to audio to remove any hissing and high-pitched sounds.
• Compression reduces the dynamic range
• The dynamic range refers to the difference in decibels between the loudest and softest sounds.
• The dynamic range should be wide enough to allow variations in expression, but narrow enough that at low volume all the sounds are still audible.
• Attack: is the amount of time it takes for the compression to ease in once the threshold has been breached.
• Release: is how long it takes for the compression to ease out once the audio drops below the threshold.
• Dramatic changes in dynamic range usually require more time to ease into and vice versa.
• Allows you to accentuate the sound without fundamentally changing it.
• Slightly nudge the mid-range or hi-range by 1 to 3dB. The effect should be a clearer audio track that sounds very similar to the original, just better
Programs like Soundbooth have lots of pre-sets. You can use these to experiment because you can listen to the effect and see in the program what has been done.
This job will always be easier with sound that is well recorded in the first place.
useful info mate.. me too a beginner in audio ... a tutorial will be very helpful
Here's the basics on how I was taught to clean audio before mixing.
Stage 1: Clean.
Scrub audio: Using your sound head listen and look for pops, pings and gargles in your audio. Use copy and and shift-command-v to copy and replace where ever find these pints.
Run noise cleaner as needed.
Stage 2: EQ
Use EQ filter to do the following:
Filter out any hum found in your audio (not always easy).
Filter out low end (40htz or below) and clip down your hiegh ends (around 20k).
Fill out the roundness of your audio. More often then naught your mic is going to either catch your high end, middle or your low end well, but never across the board. Use your EQ to boost up where your mic was deficient till you get a good round sound.
Stage 3: Boost audio levels and Compress
First, manually scrub through your audio and bring levels to match across the board (IE if you're dealing with VO's for instance, make sure they all at the same audio levels across the board).
Use a compression filter to boost the volume of your audio while keeping your peeks from db out.
Stage 4: Normalize
Use normalization filter to bring your audio down to a basic across the board range.