For the course of our final project, we alternated between three different DSP processing tools: LT Spice, MATLAB, and Logic Pro X. All three of the tools were in very different formats, with each using code, circuit diagrams (schematics), and a music production setting. As music producers and sound engineers, this project was a chance for us to explore what contemporary software practices in developing music production DSPs would look like, and we also got to discover new tools such as JUCE. 

​

Due to our lack of expertise and experience with developing ML or AI models for us to train datasets for a specific purpose, our methodology leaned toward the more traditional side. 
However, in the future, we hope to look into music datasets such as Yamaha's MAESTRO, Google's MusicCaps, or the Sound Describer Dataset to implement more complex DSP tools that would let us explore DSP techniques that are outside the traditional music production realm. Getting more familiar with JUCE would also be our future aim in exploring optimization methods for various platforms.

​

Even from the traditional method, we learned that keeping a noise-sterile environment in the data collection stage was crucial in terms of modeling a very slight frequency boost or attenuation. We also found out that MATLAB was even more useful in audio plugin generation than we previously thought, and there are so many DDSP libraries already developed for use online on platforms like Github. 

​

Overall the project was a very enjoyable learning experience and we hope to use what we have learnt here in our final senior project in EECS 452.

​

Thank you!

​

-Clayton, Tommy and Victor

​