Machine Listening

Principles of human auditory processing are used to develop computer algorithms that are able to retrieve relevant information from acoustical input signals.

Human listeners are able to retrieve information from acoustical inputs for many different purposes, e.g. for understanding speech or for recognizing a source and determining its distance. In comparison to computer algorithms that retrieve the same information, human auditory processing seems to be much more robust against the influence of background noise or reverberation. In the topic area of Machine Listening we are looking for ways to incorporate this robustness in computer algorithms inspired on human auditory processing principles. An example of information that can be retrieved is the simultaneous identification and recognition of human speakers in noisy and reverberant conditions.

Current Projects

    Computational auditory scene analysis

    Joachim Thiemann

Former Projects

Computational auditory scene analysis

    Computational auditory scene analysis

    Computational auditory scene analysis
    A beyesian classifier for estimating the ideal binary mask in noisy and reverberant environments

      Music information retrieval
      Eine Methode zur Gesangsdetektion basierend auf musikalischen Merkmalen und Merkmalen aus den Frequenzmodulationen tonaler Komponenten