0

I made some sound recordings of hornets' wing noise without a directional microphone, so I have background noises like wind, distant traffic, voices, birds, etc. in the recording, too.

When trying to isolate the sound I wanted, I realized that the typical noise reduction would also eliminate the sound I wanted to isolate. Likewise my attempts to isolate the sound with an expander, noise gate, or equalizer was not very successful.

I played the sound via headphones to my Android mobile device to analyze the spectrum (not having an equivalent tool on the PC ready), and here is how it looks (so note that the level seems rather low and there may be some additional environmental noises like PC fan or heating in the spectrum): Lower-frequency spectrum

Full spectrum

Also note that the lower part of the spectrum is a "waterfall graph", not showing "dB", but "seconds from now".

I even tried to synthesize the sound with a base frequency of roughly 230 Hz, but that sounded very synthetic.

Maybe it's all not worth the time and I should use some "free" professional recording instead.

Any advice?

3
  • If you have some coding skills you can try Granular Synthesis as a procedural audio method to simulate the sound you are after. If you have a lot of wasp recording you can isolate small sections ( < 100ms) with low noise, ambient sound, and use them to build the required FX. The small samples will also make it easier to filter out noise if the quality is very low.
    – Blindman67
    Commented Feb 16, 2023 at 1:52
  • If you're really serious about recording a hornet, catch one (safely) and record in an isolated environment. I'm weird about creatures, so make sure you return it to it's home after.
    – n00dles
    Commented Feb 17, 2023 at 19:15
  • @n00dles Those insects are quite clever, and it's not just very difficult to catch one, but also quite dangerous, as the rest of the family won't like it. Probably I'm hunting for a directional microphone.
    – U. Windl
    Commented Feb 21, 2023 at 10:37

1 Answer 1

1

I did not find a perfect solution, but here is what I did for partial success (mainly for those dealing with a similar problem):

Looping a fragment

As suggested by @Blindman67 I tried to loop a mostly clean fragment. Unfortunately that sounded very much "looped", even when appending the same block reversed.

Trying the synthesize the sound

I picked four rather clean fragments and captured the frequency spectrum. Here are two of them (#1 and #4) with the frequencies I picked for synthesis:

Annotated spectrum #1

Annotated spectrum #4

The logarithmic scale made it a bit hard for me to find the important frequencies, but rather surprisingly (to me) the sound has quite a lot of harmonics up to high frequencies. Therefore it's not surprising that any attempt to reduce the noise by applying an equalizer was not successful.

So for the first attempt (matching #1) I overlaid 16 sine waves, added some low-frequency tremolo, wahwah and reverb to make it sound more real ,but still it sounded very synthetic.

For the second attempt (matching #4) I overlaid 11 sine waves, added some low-frequency tremolo, wahwah and reverb.

The synthesis was done using Audacity; there are probably more powerful tools out there, but as an hobbyist I used just that. The spectrum plots were made with the "Frequency Analyzer" of DaVinci Resolve.

However I'm not very happy with the results; here are the spectrums:

Spectrum of synthetic sound #1

Spectrum of synthetic sound #4

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.