Performance difference between Slew filter and Parametric EQ

Hello Forumfanatic,

Thanks for attaching your project, I will test your conditions and get back to you.

How did you measure the before and after filter readings? did you use the internal level detector or any external tools?

Regards,

Harish

Harishgowtham
I measured the before- and after-filter levels using an external tool (SpectraPLUS). The test signals (sine tones at the specified frequencies) were fed into the DSP, and the output was captured through my audio interface and analyzed in SpectraPLUS. The dB values mentioned in the results were taken from SpectraPLUS measurements, not from SigmaStudio’s internal level detector.

Hello Forumfanatic,

As far as I checked, it seems to be working fine. Below is what I quickly tried.

One important note is that, I suggest you use Double precision filters since the corner frequency is lower. That being said, even single precision filter also gives me an expected result.

I used double precision for this test and recommend you to use double precision.

1) I set the input frequency to 300Hz with gain set to -10dB and set the filter's corner frequency to be 160Hz with 7dB boost.

You can see that the meter before the filter shows around -12dB (appx -13dB) which is expected and 300Hz input should be boosted by appx 1-2dB according to the frequency response, hence the output meter shows -11dB which is expected.

2) I set the input frequency to 50Hz with gain set to -10dB and set the filter's corner frequency to be 160Hz with 0dB boost. So no reduction or boost, the input should pass 'as is' (of course with change in phase).

The same conditions with 300Hz input.

You can check the results in the meter.

Instead of measuring within the same DSP, I tried to send this signal via SPDIF to another DSP and measured the level in the second DSP. It also gave me an expected results.

I also passed the signal to the DAC and looked at the analog out, it appears to be an expected output.

So, could you please try measuring the level in internal meter? if you are getting the same result as me with level detectors, then we will have to see what's happening in the measuring tool or the format etc.

Regards,

Harish

Harishgowtham

Apologies for the delay in getting back to you, and thank you for your patience.

Thank you for the detailed checks and the guidance on using double precision.

To clarify, my measurements are from the General EQ (2nd-order) block with Slew enabled, configured as a Peaking filter (fs = 48 kHz). I repeated your two cases using double precision and read levels on SigmaStudio internal meters (not SpectraLAB), but I still see materially different results with the Slew block.

For reproducibility, my exact settings are:

Block: Second Order → Double Precision → General EQ (2nd order) Slew

Mode: Peaking

fc = 160 Hz, Gain = +7 dB (Case 1) / 0 dB (Case 2)

Q (or BW): 1.41

Slew time: ~0.05 ms

Input: sine at 300 Hz and 50 Hz, source level –10 dBFS

Meters: RMS level meters placed immediately before and after the Slew block

Could you please confirm that your test used the same Slew block (Peaking) and parameters?
If you were on the standard Parametric EQ, that might explain why our results differ.

I’ve attached screenshots showing the block selection and meter placements.

Once we confirm we’re exercising the same Slew peaking filter, I’ll re-run any additional cases you suggest.

Thanks again for the support.

Hello ForumFanatic,

The attached screenshots of yours make sense to me and it shows the expected results.

Below is what I tried with the exact block and settings as you.

In the first case,

Block: Second Order → Double Precision → General EQ (2nd order) Slew

Mode: Peaking

fc = 160 Hz, Gain = 0 dB

Q (or BW): 1.41

Slew time: ~0.05 ms

Input: sine at 50 Hz, source level –10 dBFS

You can see in the below screenshot that, I marked the 50Hz line with red dot in the frequency response graph and you can see the gain reduction for 50Hz is -13dB.

Since the gain of the sine module is set for -10dB initially, the first level meter shows -12dB (it should be -13dB since the meter uses RMS value, but meters will give you a decent level of resolution but it's not so accurate) which is correct, and after it went thru the filter (which is set to apply -13dB for a 50Hz signal), now it shows -25dB which makes sense.

If I set the tone to output a 0dBFS signal, then I should get 13dB reduction (-13dB), and my meter should show around -16dB and I get it correctly.

In the second case,

Block: Second Order → Double Precision → General EQ (2nd order) Slew

Mode: Peaking

fc = 160 Hz, Gain = 7 dB

Q (or BW): 1.41

Slew time: ~0.05 ms

Input: sine at 300 Hz, source level –10 dBFS

You can see in the below screenshot that, I marked the 300Hz line with red dot in the frequency response graph and you can see the gain reduction for 300Hz is just 0dB (which means the signal will come and get multiplied by '1' and go out of the filter 'as is', so no reduction will happen).

So, a -10dB signal will come into the filter and go out as -10dB signal, the meters will show it as -13dB (now it shows -12dB which is fine)

Look at case 3 below,

I set the input to 300Hz ,0dBFS. You can see in the marked graph below that a 300Hz signal should get 7dB reduction (-7dB) and it correctly gets 7dB reduction, so the meter shows -10dB.

Look at case 4 below,

I set the input to 600Hz ,0dBFS. You can see in the marked graph below that a 600Hz signal should get 14dB reduction (-14dB) and it correctly gets 14dB reduction, so the meter shows -17dB as result.

So, it all makes sense to me, and I didn't find anything strange with this block.

So, please let me know about your thoughts on this.

Regards,

Harish

Thank you for your explanation, that helps clarify the block’s functionality.

As a follow-up, could you please share any reference or source for the Biquad General EQ 2nd order implementation used here?

We’re following RBJ’s Audio EQ Cookbook and would like to cross-check with the official source you recommend.

Hello ForumFanatic,

Here is the link to the wiki page for the filter formula. Yes, these 2nd order Biquad filter design uses RBJ's formula from the audio cookbook.

One thing to note is, if you manually calculate the coefficients for a particular filter setting and compared them with sigma studio's coefficients for the same filter setting, you may find some differences (which is mostly ignorable). The reason for that is explained here

Regards,

Harish