R/E/P Community
R/E/P => Recording - Engineering & Production => Topic started by: willibrown on November 12, 2012, 03:35:14 PM
-
(how's that for a search friendly subject?)
I have an archive of recordings. The levels are low but most have at least one or two spikes (where they hit the lavalier or laughed directly into the mic) which either clip or come lose. Some of these files are peaking at as low as -12, but they vary.
I can't do a bunch of manual process, like proper Noise Reduction, just have to do it quick and dirty. I want to chop these spikes as much as I dare and then Normalize (to probably -1db). Probably should just convert to mono as well.
I'm going to use (for now) the Audacity batch tool (File Edit Chains).
I will ask about the production side in a separate forum.
Thank you.
-
The Audacity compressor will do that - set the threshold to the level to which you want to limit the spikes. The ratio should be maximum (10:1). The compressor can also normalize to 0 dB, but I would recommend not doing that - your idea to normalize to -1 dB is better, though I usually leave 1.5 or 2 dB to be safe.
-
Hi,
Thanks for your reply.
Any thoughts on the attack time setting? (Range 1 to .1 secs)
Version 2 has a gang of other settings (most to ignore in this application)
Audacity Manual Entry for Compressor:
http://manual.audacityteam.org/man/Compressor (http://manual.audacityteam.org/man/Compressor)
Also, to have more precision in choosing the Threshold value, should I analyze for average peak? For example, I could have a talk that's mostly not hitting -18, but has some good chunks at -12, and then those spikes. (and yes, proper production prevents!:) )
Congrats on the book with Focal. May it benefit far and wide.
-
For this application I'd probably go with the fastest attack time and a quick release as well. To keep it simple, I'd try to figure out a threshold that works OK for all the files and "de-peak" them as a batch. The loudness matching would be done later. It's a little hard to give specifics without hearing the actual files, though.
-
Any decent plug with 50 us attack time should work. Apply threshold down until it does it's "wack a mole" thing. Set the release to about the time of the noise decay time. Watch out for fast release chopping of lower frequencies.
You can also do this in analog with an Aphex 651 comp. Set attack to 40 us, set release to match the recovery speed you want. It also has that exclusive high frequency expander that masks the compression effect, I've not seen that feature in a plug, yet.
-
Now we're talking production. (Should I move this thread?)
I have people who record without checking levels. Thankfully they are usually too low, rather than too hot. Putting that compressor on the output of the Mackie, okay?
Aphex has a good rep. I don't have the ear or expertise to judge that stuff. Would love an 2nd option that is new and commercially available - the application isn't demanding of high fidelity.
Thanks.
-
A mastering limiter would be a better tool. The controls and processing are more suited to this, often such as one knob/slider that sets the threshold and the makeup the gain at the same time. A second control is generally provided for putting in some headroom.
So you just pull the threshold control down 15dB (for example) which limits peaks to -15dBFS and also adds 15dB of gain, then you pull the output down to your -1dBFS target. Done. (You would probably need to do this to files one at a time rather than batch processing them, so you can measure the starting RMS value and set the threshold to get the audio to a standardized RMS.)
I'm pretty sure there are plugins out there that will normalize to a specified RMS level by limiting peaks and applying the needed gain. That might allow for easy unattended batch processing of the whole archive.
-
I don't know much about the "Audacity" program... but if its possible to go into the program and automate the "spikes" to little or no level that might be a viable method for their removal... and subsequent "normalling" of the audio.
No?
-
I don't know much about the "Audacity" program... but if its possible to go into the program and automate the "spikes" to little or no level that might be a viable method for their removal... and subsequent "normalling" of the audio.
No?
I have an archive of recordings.
I can't do a bunch of manual process...just have to do it quick and dirty.
He really needs an automatic process, preferably one that could limit and amplify as needed to reach a target RMS level.