Audio sounds similar in volume before and after filtration of the most energetic frequency band

Hello

I have an audio recording of environmental noise, about 8 sec of the recording. Processed it to reveal the frequency bands carrying the most energy and filter it out.

Used spectrogram in order to see the frequency spread over timeline of the recording, looks like the most is focused in lower band of under 100 Hz thoughout the entire recording. Attached is the spectrogram results.

Then used pwelch to figure the most energetic frequency band, showing peak and about 40 Hz. Used Butterworth order 5 HPF at cutoff of 120 Hz to suppress the low band.

The pwelch of the original recording and the filtered data is attached. Looks like shaved off the low band nicely up to about 120 Hz.

Printed out the sound data before and after the filtration - the difference in the std is obvious.

With all that, when I'm streaming the original and filtered signal to my PC speakers (using either sound() or audioplayer() functions) - it sounds idential to my ears in frequency content and in volume (of course, given the same PC volume).

I'd expect the filtered signal would sound much quieter due to the band carrying most of the energy is filtered out.

What do I miss in my understanding ?

P.S. the original WAV file is also attached just in case

Thanks to all willing to help

[Daniel M]

First check if your speakers are trash by doing a low pass filter instead of high pass (i.e. keep only the ~40 Hz signal), and listen to that sound. Can your speakers play low-end tones? Is it quieter, or lower pitch than you were expecting?

Note that the cutoff for human hearing is around 20 Hz, and that the volume threshold for hearing is a lot higher at very low frequencies than it is at higher frequencies. That is, even though the 40 Hz band has more energy, it will still be more difficult to audibly perceive than 500 Hz at a lower energy.

[Alex Z]

Thank you Daniel, spot on!

Checked my laptop speakers - they do not reproduce such low range so no any effect of the filtration is heard through the speakers. Connected my headphones - now indeed the lows can be distinguished and the filtration effect is here.

Having said that, indeed, even though the analysis shows most energy being in very low frequencies, filtering them out makes little effect in terms of listening. The roar is suppressed but it turns the most of the noise is still heard by ears. So looks like your note of human hearing volume threshold for very low frequencies is considerably higher. playes its role in my case

Need to understand better human hearing sensitivity scale first before trying to apply any processing on sound that is aimed to human ears..

Thanks you

[Daniel M]

No problem. What are you trying to accomplish anyways?

[Alex Z]

Trying to study certain aspects of active noise cancellation. For the beginning just to brush up my memories from basic signal processing by just filtration of recorder signals and making sure the results are in line with my expectations

Answer 1

[Daniel M]

This has been solved. The answer was poor speakers and human threshold of hearing stow frequencies (as per the comments above).