Achtung:

Sie haben Javascript deaktiviert!
Sie haben versucht eine Funktion zu nutzen, die nur mit Javascript möglich ist. Um sämtliche Funktionalitäten unserer Internetseite zu nutzen, aktivieren Sie bitte Javascript in Ihrem Browser.

Info-Icon This content is partly available in English

Source Separation

“Source Separation” is a wide research area ranging from decomposition of EEG signals to telecommunication channels. We will here focus on speech signals and mention the obligatory cocktail party problem:

Imagine a cocktail party where different speakers speak simultaneously. A keen listener may try to unterstand a speaker but struggles interfering speaker and noise.

Our past research projects aimed at solving this by leveraging phase and level differences in a model based clustering approach to estimate spectral masks for each target speaker. It became apparent that this approach is limited due to the lack of knowledge from an extensive database and not using speech characteristics.

A more recent outcome is a discriminatively trained neural network to provide spectral masks to estimate beamforming coefficients in a multi-channel setting. Motivated by its efficacy demonstrated during the third and fourth CHiME challenge, our current research focusses on:

  • Discriminatively trained spectral masking,
  • Synergy of model based source separation with neural networks.

Source separation and interference reduction for automatic speech recognition in dynamic acoustic environments (Transfer Project)

This project is dedicated to a holistic approach for speech enhancement, separation and recognition in an automatic house environment. Financed by the “Deutsche Forschungsgemeinschaft” (DFG) a new system is researched combining the hand-on experience from a third-party company with the recent scientific advances in speech enhancement through deep learning at the Department of Communications Engineering (EIM-E/NT) . As basis for the speech enhancement and separation task a spectral masking beamformer inspired by an approach developed in our department is considered.

Ansprechpartner

Lukas Drude

Nachrichtentechnik (NT)

Research & Teaching

Lukas Drude
Phone:
+49 5251 60-5288
Fax:
+49 5251 60-3627
Office:
P7.2.06
Web:

Jens Heitkämper

Nachrichtentechnik (NT)

Research & Teaching

Jens Heitkämper
Phone:
+49 5251 60-5288
Fax:
+49 5251 60-3627
Office:
P7.2.06

The University for the Information Society