Bird song editing steps, anomalies

Audacity 3 Help Forum Windows
1

I am using Audacity 2.3.2 on Windows 10 Enterprise (Version 1703, OS Build 15063.1805). I have been recording bird songs as WAV files using Voice Record Pro on my iPhone 6s with the end result to add them to the eBird.org database. (WAV are preferred by eBird.)

eBird’s guidelines say to “normalize” or “boost” the sound of recordings: “Adjust the level of each sound file so that the loudest bird sound reaches -3 dB.”

I think I have figured out how to do that but want confirmation from other Audacity users.

What I’m doing:

  1. Open file.
  2. Ctrl-A
  3. Effect → Amplify
  4. Set “New Peak Amplitude (dB):” to -3.0
  5. OK
  6. File → Export as WAV
  7. Save as type: WAV (Microsoft) signed 16-bit PCM (the default that comes up), SAVE.
  8. OK (on “Edit Metadata Tags” - haven’t made any changes)

Sometimes, somewhere on a track, -3.0 dB has already been exceeded. Is going through the steps going to enhance the recording or make it unnecessarily softer?

I know how to Trim files by:

  1. Place cursor at boundary. (It seems you have to be very precise as to how high on the line you click to place the cursor. Too low or high and it starts playing instead.)
  2. Select → Region → → click on Trim icon

On one recording, there was some wind when recording. It resulted in a couple of very high, but narrow spikes in the sonogram (?). Is there an easy way to manually shorten those spikes so that the normalizing isn’t affected by those? I see how to “Silence audio selection”, but ideal would be to not have a blip of complete silence. (I could attach a file of that would help.) (And if you think I should post this as a separate, new post, please let me know.)

In another recording there is a distant lawn mower noise throughout. Any easy way to get it out?
In another recording someone is talking for a couple of seconds–easy way to get it out?

Thank you!

2

Very difficult! It’s hard to get a strong signal-to-noise ratio. It wouldn’t hurt to apply a high-pass filter (or use the equalizer to reduce the low frequencies). I don’t know what the cutoff frequency would be, but birds don’t make “bass” so any low frequency sounds are just noise.

What I’m doing:

  1. Open file.
  2. Ctrl-A
  3. Effect → Amplify
  4. Set “New Peak Amplitude (dB):” to -3.0
  5. OK
  6. File → Export as WAV
  7. Save as type: WAV (Microsoft) signed 16-bit PCM (the default that comes up), SAVE.
  8. OK (on “Edit Metadata Tags” - haven’t made any changes)

Perfect! Of course, that should be the last step after any editing/processing.


Sometimes, somewhere on a track, -3.0 dB has already been exceeded. Is going through the steps going to enhance the recording or make it unnecessarily softer?

When you set a “New Peak Amplitude” Amplify Will adjust the level up or down as needed. (Down is actualyl attenuation instead of amplification.)

If you are hitting 0dB before any processing, you are clipping (distortion). That may be OK if you can’t hear the distortion. Normally I’d say lower the recording levels, but it’s probably too low most of the time and you can’t control the birds. (just for reference Pros typically record at -12 to -18dB and then amplify later. You don’t need a “hot” signal with digital recording.)

On one recording, there was some wind when recording. It resulted in a couple of very high, but narrow spikes in the sonogram (?). Is there an easy way to manually shorten those spikes so that the normalizing isn’t affected by those? I see how to “Silence audio selection”, but ideal would be to not have a blip of complete silence.

You can use the [u]Envelope Tool[/u]
to fade-down and then fade-up the volume (if that makes it less annoying/distracting). That’s the trick with the Envelope Tool… Don’t make sudden jumps in volume.

In another recording there is a distant lawn mower noise throughout. Any easy way to get it out?

Hopeless. There’s a reason on-location movie dialog is re-recorded in the (soundproof) studio. :wink: There is a Noise Reduction effect, but it works best when you have a constant, low-level, background noise… It works best when you don’t really need it. If the noise is bad, “The cure can be worse than the disease.”

In another recording someone is talking for a couple of seconds–easy way to get it out?

I won’t attempt step-by-step instructions, but I’d split it into two (or more) recordings, fade-out the 1st part, fade-in the 2nd part, and re-join with an overlap to create a smooth crossfade. The fades can be just a few milliseconds so you don’t “hear” the crossfade, but you get a smooth splice.