Creative Audio Effects for Music

Audio effects turn familiar instruments and recordings into new sounds: a guitar pedal adds character, a harmonizer creates extra voices, and a reverberator transforms a short note into an evolving texture. This student project explores how we can design expressive effects and give musicians useful control over them. Approaches can use classical digital signal processing (DSP), deep learning, or hybrid methods such as differentiable DSP.

Our group’s work offers several starting points: matching nonlinear audio effects with Koopman-based neural networks (KLANN)[1], creating shimmer reverberation with nonlinear feedback delay networks[2], and exploring zero-phase sound and phase-scaling effects using giant FFTs[3][4]. The project can build on these ideas or explore another effect of musical interest.

Possible directions

• Guitar effect matching: reproduce the sound and dynamics of a distortion pedal or amplifier from input/output recordings, using DSP models, neural networks, or a combination of both.
• Nonlinear and shimmer reverberation: explore distortion, modulation, and pitch transformations inside a reverberator’s feedback loop to create evolving textures with controllable decay and stable behavior.
• Pitch shifting and harmonizers: develop effects that transpose an instrument or add harmonically related voices, and investigate sound quality, musical control, and latency.
• Zero-phase and spectral effects: manipulate spectral phase to reshape transients and timbre, exploring zero-phase processing and extensions for creative sound design.

These are alternative starting points. We will choose one direction together and agree on a scope suitable for a research project or a master’s thesis; your own effect ideas are welcome.

Tasks

• Choose an effect and a concrete design or modeling question, then implement a baseline.
• Develop and compare a DSP-based, learning-based, or hybrid approach using suitable recordings and musical examples.
• Evaluate sound quality and controllability through listening and signal analysis; assess computational cost and latency where real-time use is a goal.
• Present a working prototype with audio demonstrations and document its strengths and limitations.

Prerequisites

Experience with digital signal processing and programming in Python, MATLAB, or C++, plus an interest in music production or sound design. Learning-based directions also require familiarity with deep learning. We can adapt the topic to your background.

References

1. Ville Huhtala, Lauri Juvela & Sebastian J. Schlecht (2024). KLANN: Linearising Long-Term Dynamics in Nonlinear Audio Effects Using Koopman Networks. IEEE Signal Processing Letters.
2. Gloria Dal Santo, Xiaojie Pi, Karolina Prawda et al. (2026). Shimmer Reverberation with Nonlinear Feedback Delay Networks. Proceedings of the 29th International Conference on Digital Audio Effects (Dafx26).
3. Vesa Välimäki, Stefan Bilbao, Sebastian J. Schlecht et al. (2025). Zero-Phase Sound via Giant FFT. International Conference on Digital Audio Effects (Dafx25).
4. Vesa Välimäki, Roope Salmi, Stefan Bilbao et al. (2026). Phase-Scaling Effects Using Giant FFT. Journal of the Audio Engineering Society.

Contact

Prof. Sebastian J. Schlecht (sebastian.schlecht@fau.de).