Proper levels in cropped field recordings

I do field recording and crop them up into smaller pieces for use in my music. Usually the field recordings end up having low volume. I would like to increase the volume/gain on them. I’m just not sure how I should do this? Each recording has its particular volume. What should be the level I should aim for and how do I judge whether I’ve attained the proper level? Is it some kind of average volume I should attain?

Or should I use a compressor for this purpose?

I can usually achieve what I want but I’m looking for a more systematic approach to this problem. Thanks for the help.

Assuming your work is neither so quiet that it’s noisy or so loud as to be distorted, I think what you’re doing is perfect. Compressors don’t just change loudness, they also, in extreme, change the density of the sound. You never get away with no evil effects. The best you can do is balance the effects versus the sound quality and time for production.

Generally, what people want is to make the sound as loud as possible, so Maximum Volume Before Clipping is an unfortunate, but well-defined goal. That’s not what you have. You have the goal of making two or more different clips match. The broadcasters do this by smashing all shows and sound packages to the same volume. In some instances, this is rigidly defined in law. This is why you generally never have to adjust the radio in your car. Unless you have a very noisy car, you can tune in a station in 2009 and never change it until you sell the car.

You can get that kind of stiff regulation with Chris’s Compressor. Chris designed his product so he could listen to opera in the car. Everything from Orchestra Tutti to one fiddle in the south forty comes out about the same volume.

I use it to prepare a completely out of control podcast (downloaded) for listening in the car. Oddly, exactly what was intended.

Give it a shot. It makes no pretense of retaining the original dynamic range of high and low volumes, so if that’s important to you, you’re back to doing it by hand.

http://theaudacitytopodcast.com/chriss-dynamic-compressor-plugin-for-audacity/

The show also gets louder. This is a sample of the same clip with no compression, default compression, and the compression setting (first slider) reset to 0.77.

http://kozco.com/tech/audacity/pix/ChrisCompressorBeforeandTwoAfters.png

Koz

Thanks for this very informative post. I am definitely not trying to join the loudness wars. I can imagine that I would want to keep the dynamics of some field recordings; maybe when there are soft sounds that gradually grow into louder ones for instance. But in those cases, I can decrease the strength of the compression.

So, I will try what you recommended and report back if necessary. Cheers!

So, I used this compressor and it’s nice that there are brief explanations for what each parameter does. However, the incoming audio differs because my field recordings are not done under standard conditions (even if I keep the input gain the same, the nature of what is recorded varies widely of course). Thus, I can kind of stick to the same parameters but this would still create variable results. How can I be more precise about standardizing my various field recordings? I want to make sure they are at an optimal level and close to each other so that I don’t have to think about this once I process them with Audacity and throw them into my DAW or sample manglers for further use.

How can I be more precise about standardizing my various field recordings?

Record them in a studio. By definition, Field Recordings are created under uncontrolled and unstable conditions. Very few people collect Field Recordings, throw them on a timeline and go without post-production and balancing.

Some field sound mixers have sound processing built-in. My Shure FP33 will do that with a LIM (limiter) switch on the front.

This circuit keeps the maximum peak output level below the level set by the user and prevents clipping…

This means you can ride the show level higher and more consistently without worrying about snapping and crunchy sound from overload. The downside is accidentally riding the show too high resulting in obvious Pumping and Recovery effects.

Depending on your work, that may be OK. Straight vocal interviews can work pretty well that way.

~~

You get the impression I have no idea how you’re capturing the work or even what the work is. A humming bird shoot is very different from capturing dialog in a speedboat on Biscayne Bay (actual shoot a friend of mine did). Describe what you’re doing.

I get the fuzzy feeling you are trying to do field recording by setting a recorder once in the morning and using it that way all day long without looking at it. It doesn’t work that way. Field recorders have volume meters for a reason. They’re not optional.

This is one of my mixers going out on a shoot. It qualifies as a successful product. There’s no paint left, you can’t read the knobs any more and you can’t have it right now because it’s out on a job.

Which mixer/recorder do you use?

Koz

Thanks. I don’t just set my recorder and forget about it, but neither am I trying to establish myself as the next big name in field recording. I have a Zoom H1 and I can set the input gain well enough that I’ve never had a clipping problem. I record various things, I’m open, anything is suitable for use in my music. Mostly ambient urban settings. People talking, traffic, etc.

I’m not trying to establish precision at the recording stage. I’m trying to establish a way to generally get my recordings to about the same level has each other. What is this proper level roughly, how do I measure it in Audacity (i.e., is it an average, what’s the unit, which Audacity function shows it), and how do manage the gain on my recordings so they are about the same healthy level and ready to be fed into my sample manglers and DAW? These are my main questions right now.

(i.e., is it an average, what’s the unit, which Audacity function shows it)

You would think this would be a button push or simple set of tools, but it isn’t. Most Audacity tools are simpletons. Like Amplify and Normalize, they work on peaks in the blue waves under the assumption that most of the show is going to follow along. It doesn’t have to. I know of no simple or easy tool that works directly on overall clip loudness. The blue waves and bouncing sound meter are peak-reading. They ride and respond to maximum loudnesses, not overall average. The silly joke is of a gunshot or dropped saucepan in a clip is enough to throw all the tools far enough off and destroy the rest of the clip, and it doesn’t matter how loud everything else is. The clip loudness will be hijacked by the saucepan.

It’s possible we can derive something, but it would be neither easy nor convenient.

(simplified narrative)
Import the sound clip and apply the A loudness contour. The clip now will ignore everything you can’t easily hear. Play the clip and look at the light blue bands inside the blue waves on the timeline (attachment). Those bands should indicate the loudness of the clip. You can then change the clip up or down (Amplify or Normalize in manual mode) to match a pre-determined standard.

Then it gets magic. Write down the gain or loss required and apply it to a copy of the original clip, making sure there is no overload damage.

This is a similar problem to the requirements of audiobook reading. No overly theatrical volume changes. That’s not easy, either.

Koz
RMSWaves.jpg

You can do this with the bouncing light sound meters, too. The color change about 2/3 of the way up should be the same value as the light and dark blue bands except in real time. You should make the sound meters very much larger for this kind of work. Click the right-hand edge and pull sideways.

http://kozco.com/tech/audacity/Audacity1_playback.jpg

But you still have to get the A Curve in there for best accuracy. That means the system won’t respond to a bus driving by rumble or overly high-pitched sounds. They’re not theatrically significant, but Audacity can’t tell.

Somebody built that curve a while ago…

Koz

This is from another thread:

“A Weighing” takes into account how your ear works.
You are warned against audio processing using RMS values. You find very rapidly that bass notes which have a terrifically large RMS value take over the processing and produce annoying pumping effects during music.

This is why you need the A Weighing in addition to the color changes on the timeline or the sound meters to see how loud you are.

I know somebody wrote it and posted it. I just can’t find it.

Koz

I did a quick interpretation of the A Weight Curve (attached).

Download it to your computer, open Audacity with any sound file. Effect > Equalization… > Save/Manage Curves > Import. Choose that file. That should give you an equalization curve like the attached picture. Change the “length” of the filter to maximum.

If you apply that equalization to a clip, all the unimportant high and low pitched sounds should vanish leaving just the important stuff in the midrange. The light blue timeline waves or the lower of the two bouncing sound meter colors should then indicate the approximate real-world loudness of your clip.

Remember, you asked for this.

Koz
Screen Shot 2014-10-29 at 16.49.13.png
A-Weight.xml (709 Bytes)

Choose a clip whose loudness you like. That can be your standard clip.

Effect > Equalization: A-Weight, Length Maximum.

Note where the light blue sections of the blue waves on the timeline are or where the lower of the two colors on the sound meters are. This is your standard. Make a screen capture so you remember it. Do Not Save or Export Anything. Close Audacity.

Open a copy of your Field Recording. Never do effects or filters on original live captures.
Effect > Equalization: A-Weight, Length Maximum.

Note where the new light-blue waves are or the lower of the two sound meter colors bounce.
If they’re not the same as your standard, note how much lower or higher they are and Effect > Amplify: Amplification (dB)[make-up value] > OK.

Check that it worked. Check that the new value of loudness meets your standard. Edit > UNDO if you need to adjust it.

Close everything and open a fresh copy of the clip. Amplify: Amplification (dB) [the final makeup value you picked] > OK.

Your standard clip and the newly patched clip should match in overall loudness. Export to your system.

Koz

I did the last thing, but I don’t completely understand what I’m doing and it seems that it comes at the cost of altering the recording in a way that I do not want. I understand with your replies that this issue is beyond my grasp. I think I can do better just applying standard compression to all my recordings. That should do the trick.

Thank you very much for you help. Much appreciated.