Dr. Youssef El Baba

Tasked with algorithm development for acoustics-based, regulated medical sleep analysis.. Work focused on conceiving and implementing an algorithm and processing for sleep apnea detection using microphone signals (“Snorefox”), and scientific accompaniment and validation. I developed an algorithm and implemented it on my own from A to Z, conceived a solution for extracting a bio- from an audio-signal, and invented a scheme for post-processing results (patent pending), resulting in high accuracy / robustness.

Tasked with continuing research on acoustic room geometry inference, where I conceived a creative & efficient solution for wavefront detection (the first in the field for this NP-Hard problem). Also did research in electroacoustics (study of properties of microphones and loudspeakers) and low-frequency (modal) room acoustic behavior, where I invented a method for acoustic surface property (impedance) estimation. Resulted in 2 published journal papers and 3 conference papers. PhD diploma and patent obtained.

Studied the influence of positioning restrictions in mic and speaker measurements and investigated known approaches, while performing acoustic measurements to validate the chosen approach for inferring room characteristics from room impulse responses (RIRs).

Researched image quality metrics to implement a blind such method, and statistically detected a “sifting” threshold used to predict decodability. Also researched image super-resolution, used that knowledge to develop a custom image fusion algorithm for the company’s applications, and then went on to research blind blur deconvolution and developed an in-house solution, achieving 60% quantitative improvement on treated images. Other work areas included image texture analysis and recognition

Have been working as an assistant in managing EPFL's Digital Signal Processing Massive Open Online Course (MOOC) 's Forum, given by Pr. Paolo Prandoni & Pr. Martin Vetterli. Task is to answer questions on the forums about DSP theory, the given online lectures and the homeworks.

Ensuring that the Montreux Jazz Festival Archive's concert files are of proper quality: meausuring basic audio and video characteristics (max RMS, peak, average peak etc) and detecting potential problems like dropout, bad chrominance/luminance and color phase inversion.

After having finished my internship at Richemont during summer 2012, I was offered the same position again for summer 2013. Mainly worked on reorganizing the email infrastructure for the company and presenting the available resources in a clear, understandable way, enabling the CRM team to instantly spot weak & outdated/unused links.

I helped the Richemont SAP CRM team test new products coming out of development, and insure software quality. I've also relieved the team of many routine tasks, greatly improving its efficiency and productivity. I can confidently assert that I was productive since the very first few days, which is important in the case of a short term internship.

Specialization in signal processing