2005 - 2008: Technical University of Denmark
Then main topic of the study is separation of speech and audio using non-stationary mixing models:

• Development of new non-stationary mixing models.
• Parsimonious parameterization of the mixing matrix using semi-physical modeling of e.g. the room echo response for acoustical signal separation.
• Online estimation algorithms.
• Separation based on few signals, preferably one or two signals.
• Incorporation of domain specific source signal models with inspiration from related research areas such as speech processing, coding and  ecognition, computational auditory scene analysis (CASA).

It is investigated how knowledge of the characteristics of the source signal and mixing matrix can be used to solve the source separation problem. A signal model can e.g. include inter-frequency correlation, inter-frequency coherency, and geometrical constraints.

2001 - 2003: Aalborg University
Specialization: Speech Communication, Signal Processing
Project work includes:

• Masters Thesis: Perceptual Unitary ESPRIT Algorithm
  Estimation of perceptual relevant model parameters for sinusoidal audio coders by incorporating a psycho acoustic model in the unitary ESPRIT algorithm.
  Key topics: Subspace based signal processing, Speech and audio coding, Human auditory perception, Linear algebra.
  Grade: 11
• 9th semester: Voice Triggered Drums
  Real time voice triggering of drums based on artificial neural networks and hidden Markov models.
  Key topics: Speech recognition, Real time systems, Statistical estimation, Neural networks, Hidden markov models.
• 8th semester: Implementational Studies of Adaptive Lattice Filters
  Study of efficient hardware implementations of adaptive lattice filters using coordinate rotation digital computer (CORDIC) algorithms.
  Key topics: DSP processor architecture, ASIC, VHDL.
• 7th semester: Relation between channel count and quality of sound equalization
  Investigations on how the number of employed audio channels influence the equalization of a low frequency sound test chamber.
  Key topics: Room equalization, Optimal filtering.

1998 - 2001: Herning Institute of Business Administration and Technology
Grade Point Average: 11.2
Project work includes:

• Bachelors thesis: Diffuse Field Generator
  Development of a system generating a diffuse field of pink noise using several loudspeakers in a sound test room for testing wireless phones. The project was carried out in cooperation with RTX Telecom.
  Key topics: Room equalization, Optimal filtering.
  Grade: 11
• 5th semester: Software for PC and EZ-Kit Lite
  Development of software for the PC and a DSP-processor board from Analog Devices that enables real-time logging and visualisation of data.
  Key topics: Real time systems, DSP hardware.

1994 - 1997: CEU Herning
Grade Point Average: 10.6

 

Mikkel N. Schmidt | Technical University of Denmark | Email: mns(a)imm.dtu.dk