Would it be a good idea for font makers to low pass filter all sounds at minimum 200 Hz?

Frequency response charts show what the real-terms output of the speaker is when playing a frequency at a constant volume. If you modulated a constant sine wave output moving up the range, you won’t get the same output ‘volume’ consistently. For example, in ear monitors might have multiple drivers built in, and there won’t be equal crossover between them, causing a dip in those crossover frequencies.

With these single-driver speakers, I’d guess the chart would be mostly solid across the response range, but drop-off at the top and bottom ends would be quick and that bass drops off before the sub-bass ranges. There may even be some dip and then a blip in the response due to materials resonance.

Sunday I’m free, I’ll try to squeeze in some testing. My original plan was to plot freq response curves. Multiple speakers, same chassis and hilt, a room EQ mic and software, with pink noise in a mild dampening surround setup.
The speakers I have are VECO, KR, Smuggler’s, TCS WOW, and a variety of AliExpress ones, which are likely the same as some of the aforementioned. We’ll see.

Ultimately, all we need to know is at what frequency these speakers turn into resistors rather than speakers. :slight_smile: If possible, it’s probably best to measure at a realistic distance, say 2 feet or so. If we measure too close, the back-wave cancellation effect might not be taken into account.

1 Like

I did some filter testing today. So far, my recommendation is to not use the filters, here is why:

I generated a logarithmic 60-second sine sweep:

(Note that it doesn’t sound like it’s playing for the first 10-30 seconds, because it starts at very low frequencies.)

The spectrogram of this sound should ideally look like this:

When I play it on my computer speakers, and record it through a cheap reference microphone, I get this:

(Note that there is some background noise, so ignore anything that isn’t a diagonal line…)

When I play it on a proffieboard, using volume = 1000 and a small 10w speaker I use for bench testing, I get this:

I then tried it with a 100Hz cutoff filter, order = 4:

See all those overtones? Not only are those overtones audible, but they create distortions as well. I tried a bunch of different filter settings, and they all have these overtones. I’m not sure if these are normal for butterworth filters, or if my filter implementation is buggy in some way. Either way, it’s difficult to recommend using filters based on this.

I should also mention that these spectrograms are more sensetive than my ears. The 10w speaker might look like it can play 70Hz sounds, but it’s not until around 100Hz that my ears can hear it.

1 Like

Pretty sure that’s what was registering here too? These are octaves.

1 Like

Is this a programming issue? I wonder if it’s causing resonance when it filters?

1 Like

Not sure. My next step is to run the waveform through an external butterworth filter and see if it behaves the same way.

Initial tests of the filtering code seems to suggest that it’s working quite well.
However, it seems to cause some amplifications at some frequencies, which is not expected, and if that amplification leads to clipping, then that might explain the overtones we’re seeing.

1 Like

It looks like the instructions I was following for how to create filters had some steps wrong. Hopefully it will work better once I’ve straightened this out. :slight_smile:

3 Likes

Is this getting delayed from 6.x release?

No, I just need to fix it.

I think I have the filtering sorted out now.
Assuming it works, I should be able to get back to speaker testing.

3 Likes

Ok, I think I have it working properly now.
Here is a standard 28mm TCSS “bass” speaker playing the sweep.

Like before, the micriphone is more sensitive than my ears. I can feel the speaker vibrating, but I can’t hear anything until it gets to about 150 Hz.

So, here it is with a 130Hz, 8th-order filter:

I can’t seem to tell any difference by ear, but the speaker definitely vibrates less at lower frequencies.

I think for this speaker, the recommended cut-off frequency should be somewhere in the 120-150 Hz range. An 8th-order filters seems to work well, but we might want to do more testing to figure out what is actually best.

3 Likes

Fantastic! Have these changes been implemented in 6.1?

Yes, it has.

3 Likes

You need to work and test with the WOWsound TCS 28mm 3W (4W peak) bass speakers, or the Smugglers Outpost ones, but not the “TCSS standard 2W VECO bass speakers”, because they have the worst sound quality.

• TCS WOWsound one (best sound clarity on the market for a 28mm bass speaker): https://www.thecustomsabershop.com/4W-28mm-WOWSpeaker-P1511.aspx
or from the original supplier:
https://tcsdcc.com/1694

• SO bass speaker:
www.etsy.com/listing/830376131