Removing Background Sound?

[brispuss]

I have an audio clip of an engine running at around idle speed. The problem is that there seems to be some slight background rhythmic sound which I believe shouldn't be there.

The main engine sound seems OK, but there is a background noise (sounds a bit like shwar shwar shwar shwar etc) which is distracting and shouldn't be there, or at least not noticeable.

How to remove the background sound from the enclosed audio clip as far as possible without effecting the main engine sound using an audio editor?

[cholla]

Sounds like the bearing in something.
The alternator for example.
Remove the drive belts or serpentine belt.
Then start the engine & run short term see if the noise disappears.
If it does then it is something the drive belts are running.
If not it is some other bearing or bushing noise.
For example if your radiator is cooled with electric fans.
If you have a manual shift transmission it could be the throwout bearing.
Too many bearings can make a similar noise to make an online diagnosis.

[brispuss]

Thanks for the comments.

Sorry, but I didn't make myself clear enough. I'm wanting to remove the background sound from this audio clip using an audio editor.

I've amended my first post to make it clearer.

[davexnet]

Do you mean this sound in the low frequencies?

[brispuss]

No.

If you play back my engine.wav, and especially when you loop the playback, you can hear a sort of low to mid frequency sound like shwaa shwaa shwaa behind the engine idling sound. That shwaa sound is the sound I would like to remove from engine.wav.

[timtape]

Below is a still from a moving sound image. Imagine a cursor moving from left to right, following the sound as it plays.

Is this the rhythmic sound you mean? There are about three instances per two seconds. It's a sound regularly rising and falling in frequency. It's not easy to hear or even see the sound in the spectrograph without repeated listening and focussing on that particular sound.

Assuming I've correctly identified the sound, I must ask what is your reason in wanting to remove it? The reason I ask is because the sound you seem to want to remove is itself already very weak and is mostly masked by the other stronger sounds. Removing it, assuming it were possible, would only allow us to hear a tiny bit clearer the other sounds which we can already hear quite well.

[timtape]

Sorry, double posted. Couldnt work out how to delete. Tim

[Donald Mack]

To remove background noise from your engine recording, use Audacity's Noise Reduction tool. Select a noise-only section, get the noise profile, and then apply noise reduction to the entire clip with mild settings (12-20 dB, Sensitivity: 6). Optionally, use a high-pass filter to remove low-end rumble. In Adobe Audition, capture a noise print and apply noise reduction to the full clip, adjusting settings carefully to preserve the engine sound.

[brispuss]

Below is a still from a moving sound image. Imagine a cursor moving from left to right, following the sound as it plays.

Is this the rhythmic sound you mean? There are about three instances per two seconds. It's a sound regularly rising and falling in frequency. It's not easy to hear or even see the sound in the spectrograph without repeated listening and focussing on that particular sound.

Assuming I've correctly identified the sound, I must ask what is your reason in wanting to remove it? The reason I ask is because the sound you seem to want to remove is itself already very weak and is mostly masked by the other stronger sounds. Removing it, assuming it were possible, would only allow us to hear a tiny bit clearer the other sounds which we can already hear quite well.

Yes, the nuisance sound (as I hear it) is not particularly noticeable, but I can definitely hear it. That sound just doesn't seem right and shouldn't really be present I believe. So I would prefer to remove it altogether (if possible).

[brispuss]

To remove background noise from your engine recording, use Audacity's Noise Reduction tool. Select a noise-only section, get the noise profile, and then apply noise reduction to the entire clip with mild settings (12-20 dB, Sensitivity: 6). Optionally, use a high-pass filter to remove low-end rumble. In Adobe Audition, capture a noise print and apply noise reduction to the full clip, adjusting settings carefully to preserve the engine sound.

Thank you for your very helpful comment.

I will look into these noise reduction processes when I have a bit more spare time (busy with many other things right now).

[davexnet]

To remove background noise from your engine recording, use Audacity's Noise Reduction tool. Select a noise-only section, get the noise profile, and then apply noise reduction to the entire clip with mild settings (12-20 dB, Sensitivity: 6). Optionally, use a high-pass filter to remove low-end rumble. In Adobe Audition, capture a noise print and apply noise reduction to the full clip, adjusting settings carefully to preserve the engine sound.

Thank you for your very helpful comment.

I will look into these noise reduction processes when I have a bit more spare time (busy with many other things right now).

This type of noise reduction will not work. There is no clear section to get a noise print from

[BW37]

I can’t help with modifying the wav file, but as a “car guy” I’m guessing that the “nuisance sound” is something like a drive belt with a frayed edge or similar that is dragging on the edge of a pulley as it cycles around its path. I say this because the sound seems to be just as repetitive as the main pulsing of the other engine sounds and is surprisingly, nearly in phase with them in the short term (as opposed to a longer term phased beat).

It may be something you’d rather not have in the sound file, but I think it a real sound made by something cycling proportionally with the engine’s rotation.

BW

[timtape]

Below is a still from a moving sound image. Imagine a cursor moving from left to right, following the sound as it plays.

Is this the rhythmic sound you mean? There are about three instances per two seconds. It's a sound regularly rising and falling in frequency. It's not easy to hear or even see the sound in the spectrograph without repeated listening and focussing on that particular sound.

Assuming I've correctly identified the sound, I must ask what is your reason in wanting to remove it? The reason I ask is because the sound you seem to want to remove is itself already very weak and is mostly masked by the other stronger sounds. Removing it, assuming it were possible, would only allow us to hear a tiny bit clearer the other sounds which we can already hear quite well.

Yes, the nuisance sound (as I hear it) is not particularly noticeable, but I can definitely hear it. That sound just doesn't seem right and shouldn't really be present I believe. So I would prefer to remove it altogether (if possible).

There are various sounds going on in this clip. I think if we're going to get anywhere with this we must first agree on which sound you mean. To help narrow it down, I'm asking how many times do you hear this repetitive sound in this short clip you've provided?

It may not be your purpose to diagnose the cause of the noise but being a car guy like BW37, I think he may be onto the cause. I suspect from its rate of repetition it may point to the sound of a long serpentine belt which is perhaps damaged. Being a lot longer than the circumference of the crankshaft pulley, the belt's repetition rate will be a lot slower than the crankshaft pulley's RPM.

[brispuss]

The engine sound is actually from a large diesel prime mover within a locomotive.

The low frequency schwaa sound occurs nearly 7 times within the sample audio clip.

[BW37]

OK, not a car engine
So not likely a rubbing belt...
It still seems to be phased with the engine rhythm so to me anyway, it's not likely to be a random background sound.
Possibly random, but not likely...

In terms of trying to isolate the unwanted sound, could one suppress the louder, I think, lower frequency sounds to see if the schwaa sound can then be heard more clearly? I don't have the tools to try but just a suggestion to help us agree on what the unwanted sound is?

BW

[timtape]

The engine sound is actually from a large diesel prime mover within a locomotive.

The low frequency schwaa sound occurs nearly 7 times within the sample audio clip.

Thanks, then it seems to be the one I circled in the spectral display.

A tool I know of to try and remove the sound is found in the popular Izotope RX Advanced package using the Spectral Editor.
Here's a YT demonstration video. https://youtu.be/e9gG6JHsF6E?si=0L4WxRIfbbxBrti-
RX Advanced is available as a free but time limited download. Maybe 10 days.

[Dooby]

Firstly, I'm not an engine aficionado. They all sound like engines to me, so listening for a regular sound within an overall racket I find difficult.
These pulses (see image) are regular interjections to the overall noise. Can anyone confirm this is the problematic 'noise' or is this something else?


[Attachment 85661 - Click to enlarge]

The strength of these pulses varies, so it might be some kind of reflection.

[timtape]

Well I guess it's up to the OP to state what is the problematic sound as it's his recording.

My spectrograph was adjusted for brightness and contrast to maximise clarity of the dark swooping pattern. Unfortunately this meant those vertical lines your picture shows harder to see.

Then while your display made the vertical lines easier to see than in mine, it made the swooping repetitive sound (below about 2 kHz) very hard to see. Partly because the pattern has been squashed into a very narrow band. Ideally we need a diagram which does justice to both as I see the two as related. Some of the vertical lines I hear as a click which synchronises with the repetitive sound which occurs 6 or 7 times.

[Dooby]

Thanks, timtape. Yes, we need the OP opinion on this to confirm.
If we can identify the areas they find are out of place, then we can massage them.
It's what they hear that counts.