How to prevent distortion when applying effects (

Software Development
1

I am making a DIY guitar pedal, i made a circuit with the pcm3060 and i am using the stm32h7 as the mcu

i can get clean sound and it works just great by assigning the input to the output. but i noticed that when i try to add effects to it the signal just gets distorted. a lot.

any ideas on how to prevent this

#include "tremolo.h"
#include <math.h>

#define TWO_PI (2.0f * 3.1415926535f)
#define MIN_RATE 0.1f
#define MAX_RATE 10.0f
#define MIN_DEPTH 0.0f
#define MAX_DEPTH 0.95f  // Leave some headroom

void Tremolo_Init(Tremolo* trem, uint32_t sample_rate) 
{
    trem->rate_hz = 2.0f;  // Default 2Hz
    trem->depth = 0.5f;    // Default 50% depth
    trem->phase = 0.0f;
    trem->sample_rate = sample_rate;
    trem->phase_inc = TWO_PI * trem->rate_hz / (float)sample_rate;
    trem->smoothed_lfo = 1.0f;  // Start at unity gain
}

void Tremolo_SetRate(Tremolo* trem, float rate_hz) 
{
    // Constrain rate and convert to phase increment
    trem->rate_hz = (rate_hz < MIN_RATE) ? MIN_RATE : 
                   ((rate_hz > MAX_RATE) ? MAX_RATE : rate_hz);
    trem->phase_inc = TWO_PI * trem->rate_hz / (float)trem->sample_rate;
}

void Tremolo_SetDepth(Tremolo* trem, float depth) 
{
    // Constrain depth to safe range
    trem->depth = (depth < MIN_DEPTH) ? MIN_DEPTH : 
                 ((depth > MAX_DEPTH) ? MAX_DEPTH : depth);
}

float Tremolo_Process(Tremolo* trem, float input) 
{
    // Generate triangle wave LFO (smoother than sine for tremolo)
    float lfo_pos = fmodf(trem->phase / TWO_PI, 1.0f);
    float raw_lfo = 1.0f - trem->depth * (1.0f - fabsf(2.0f * lfo_pos - 1.0f));
    
    // Smooth LFO transitions to prevent clicks
    trem->smoothed_lfo = 0.25f * raw_lfo + 0.75f * trem->smoothed_lfo;
    
    // Update phase
    trem->phase += trem->phase_inc;
    if (trem->phase > TWO_PI * 4.0f) {  // Keep phase bounded
        trem->phase -= TWO_PI * 4.0f;
    }
    
    // Apply modulation with 10% minimum gain to avoid silence
    return input * (0.1f + 0.8f * trem->smoothed_lfo);
}

this is what the tremolo code would look like and i would apply it like this

void GuitarPedal_ProcessAudio() {
    if(tremolo_enabled) {
        float in_sample = (float)input[i].left / 8388608.0f; // 24-bit to float
        float out_sample = Tremolo_Process(&tremolo, in_sample);
        output[i].left = (int32_t)(out_sample * 8388608.0f);
    }
}

but i always get distorted sound like very distorted, not pleasant to hear

2

The samples in Daisy are floats in the range -1.0f to +1.0f .

Oops, I remember you’re not using Daisy.
Nevermind.

I wonder if your conversions between float and int are correct, I’d suggest you look at libDaisy/src/daisy_core.h

3

its working now!!! it was indeed because of the conversions between float/int and int/floats
thanks a lot

1 Like