Background

Methodology

Guitar Effects

    Tremolo

    Distortion

    Wah Wah

    Chorusing

    Delay

    Reverb       

Pitch Scaling

   Introduction

   STFT

   SOLA

   DSP SOLA

   SOLA Results

   Conclusions

 

Introduction

The second part of this project involves performing real time pitch scaling. Two methods to perform this process have been investigated. The first was using the Fast Fourier Transform, but was found to be unsuitable for DSP implementation. Therefore, a time domain approach titled ‘Synchronous Overlap and Add’ was refined to work well with guitar signals. The SOLA design and implementation is covered in chapter 5. 

A special approach is needed to scale audio signals with a high level of quality. It is not possible to simply change the sample rate to achieve true scaling. The pitch will be altered, but the length of the signal will also be distorted.  

Simple linear frequency shifting also does not give the results required. Figure 19 shows how each of the frequency components are shifted up in frequency, each by a constant value. The fundamental at 100Hz is moved to 200Hz. Each of the harmonic overtones is also shifted up by a constant of 100Hz. This process does not preserve the harmonic relationships between the frequency components. In the shifted version of Figure 19, the component at 400Hz is no longer three times greater than the fundamental. 

Figure 19: Frequency shifting by a constant 

For a signal to be scaled correctly, the frequency components need to be shifted in frequency with a logarithmic scale, rather than a linear scale. This ensures that the harmonic relationships are upheld.  

Previous | Next