Using FIR Filters

Audio Plugins
1

Hey,

I decided to have some fun and write an FIR filter. Overall, here’s what I came up with:

#pragma once

#include <JuceHeader.h>

namespace Filter
{
    juce::dsp::FIR::Coefficients<float> createLowFilterFIR(float sampleRate, float frequency, float gain, bool isShelf)
    {
        int numTaps = 101; // count FIR coeffs
        std::vector<float> taps(numTaps, 0.0f);

        float omega = juce::MathConstants<float>::twoPi * frequency / sampleRate;
        float alpha = std::sin(omega) / (2.0f * (isShelf ? 0.71f : 1.0f)); // Q for Shelf / Bell

        for (int i = 0; i < numTaps; ++i)
        {
            float window = 0.54f - 0.46f * std::cos(juce::MathConstants<float>::twoPi * i / (numTaps - 1));
            
            if (isShelf)
            {
                taps[i] = window * (std::sin(omega * i) / (i + 1.0f) * alpha); // Shelf
            }
            else
            {
                taps[i] = window * (std::sin(omega * i) / (i + 1.0f) * alpha); // Bell
            }
        }
        
        // Normalization coeffs
        float sum = 0.0f;
        for (float tap : taps)
            sum += tap;
        
        for (float& tap : taps)
            tap /= sum;

        return juce::dsp::FIR::Coefficients<float>(taps.data(), taps.size());
    }
}

Using in dsp main:

void EQAudioProcessor::updateFilters()
{
    auto LF_Coefficients = Filter::createLowFilterFIR(getSampleRate(), LF_Frequency, juce::Decibels::decibelsToGain(LF_Volume), false);
    
    leftChain.get<ChainPositions::LowCut>().coefficients = LF_Coefficients;
    rightChain.get<ChainPositions::LowCut>().coefficients = LF_Coefficients;
 
}

in main head:

    void updateFilters();
            
    using FIRFilter = juce::dsp::FIR::Filter<float>;
    using MonoChain = juce::dsp::ProcessorChain<FIRFilter, FIRFilter, FIRFilter, FIRFilter>;

    MonoChain leftChain, rightChain;

So the plugin validates fine, but when I add it to the insert, the sound disappears, and there are no errors or anything like that in the code. If anyone has developed similar filters before, I’d appreciate a few tips on how to fix the situation or investigate using DBG.