Webbläsaren som du använder stöds inte av denna webbplats. Alla versioner av Internet Explorer stöds inte längre, av oss eller Microsoft (läs mer här: * https://www.microsoft.com/en-us/microsoft-365/windows/end-of-ie-support).

Var god och använd en modern webbläsare för att ta del av denna webbplats, som t.ex. nyaste versioner av Edge, Chrome, Firefox eller Safari osv.

Online Blind Speech Extraction based on a Locally Quadratic Kurtosis Criteria and a Preprocessing Automatic Gain Controller

In English

Författare

  • Benny Sällberg
  • Nedelko Grbic
  • Ingvar Claesson

Summary, in English

This paper focuses on realtime speech extraction using blind adaptive beamforming. The speech extraction is carried out using an approximation of the kurtosis measure in a subband domain. The introduced kurtosis approximation is an improvement of a recently proposed approximation technique where a locally quadratic criterion was solved at each iteration. The improvement introduced in this paper regards an approach to normalize this same criterion using a pre-processing automatic gain control unit, and thereby making the algorithm invariant to input signal scales. The proposed method outperforms the recent technique in terms of signal to interference ratio improvement. In addition, the increased memory consumption and processing load due to the proposed improvement is comparably low and this is often desirable in a realtime digital signal processor (DSP) implementation. Further, a real-time implementation of the method is conducted and results with real data is presented.

Publiceringsår

2007-09

Språk

Engelska

Sidor

139-142

Publikation/Tidskrift/Serie

IEEE ELMAR 2007

Dokumenttyp

Konferensbidrag

Ämne

  • Signal Processing

Conference name

ELMAR-2007

Conference date

2007-09-12 - 2007-09-14

Conference place

Zadar, Croatia

Status

Published

ISBN/ISSN/Övrigt

  • ISBN: 978-953-7044-05-3