Generalization of Supervised Learning for Binary Mask Estimation

Tobias May; Timo Gerkmann

doi:10.1109/IWAENC.2014.6953357

Generalization of Supervised Learning for Binary Mask Estimation

Tobias May, Timo Gerkmann

University of Oldenburg

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

1 Downloads (Pure)

Abstract

This paper addresses the problem of speech segregation by es- timating the ideal binary mask (IBM) from noisy speech. Two methods will be compared, one supervised learning approach that incorporates a priori knowledge about the feature distri- bution observed during training. The second method solely relies on a frame-based speech presence probability (SPP) es- timation, and therefore, does not depend on the acoustic con- dition seen during training. We investigate the influence of mismatches between the acoustic conditions used for training and testing on the IBM estimation performance and discuss the advantages of both approaches.

Original language	English
Title of host publication	Proceedings of IWAENC 2014
Number of pages	5
Publisher	IEEE
Publication date	2014
ISBN (Print)	9781479968084
DOIs	https://doi.org/10.1109/IWAENC.2014.6953357
Publication status	Published - 2014
Event	The International Workshop on Acoustic Signal Enhancement - Juan les Pins, France Duration: 8 Sept 2014 → 11 Sept 2014

Conference

Conference	The International Workshop on Acoustic Signal Enhancement
Country/Territory	France
City	Juan les Pins
Period	08/09/2014 → 11/09/2014

Keywords

Ideal binary mask
Speech segregation
Generalization
Speech presence probability

Access to Document

10.1109/IWAENC.2014.6953357

OpenUrl availability

Full text

Cite this

@inproceedings{9bedf8de6adb4b57a4d434dfac72d63e,

title = "Generalization of Supervised Learning for Binary Mask Estimation",

abstract = "This paper addresses the problem of speech segregation by es- timating the ideal binary mask (IBM) from noisy speech. Two methods will be compared, one supervised learning approach that incorporates a priori knowledge about the feature distri- bution observed during training. The second method solely relies on a frame-based speech presence probability (SPP) es- timation, and therefore, does not depend on the acoustic con- dition seen during training. We investigate the influence of mismatches between the acoustic conditions used for training and testing on the IBM estimation performance and discuss the advantages of both approaches.",

keywords = "Ideal binary mask, Speech segregation, Generalization, Speech presence probability",

author = "Tobias May and Timo Gerkmann",

year = "2014",

doi = "10.1109/IWAENC.2014.6953357",

language = "English",

isbn = "9781479968084",

booktitle = "Proceedings of IWAENC 2014",

publisher = "IEEE",

address = "United States",

note = "The International Workshop on Acoustic Signal Enhancement, IWAENC 2014 ; Conference date: 08-09-2014 Through 11-09-2014",

}

TY - GEN

T1 - Generalization of Supervised Learning for Binary Mask Estimation

AU - May, Tobias

AU - Gerkmann, Timo

PY - 2014

Y1 - 2014

N2 - This paper addresses the problem of speech segregation by es- timating the ideal binary mask (IBM) from noisy speech. Two methods will be compared, one supervised learning approach that incorporates a priori knowledge about the feature distri- bution observed during training. The second method solely relies on a frame-based speech presence probability (SPP) es- timation, and therefore, does not depend on the acoustic con- dition seen during training. We investigate the influence of mismatches between the acoustic conditions used for training and testing on the IBM estimation performance and discuss the advantages of both approaches.

AB - This paper addresses the problem of speech segregation by es- timating the ideal binary mask (IBM) from noisy speech. Two methods will be compared, one supervised learning approach that incorporates a priori knowledge about the feature distri- bution observed during training. The second method solely relies on a frame-based speech presence probability (SPP) es- timation, and therefore, does not depend on the acoustic con- dition seen during training. We investigate the influence of mismatches between the acoustic conditions used for training and testing on the IBM estimation performance and discuss the advantages of both approaches.

KW - Ideal binary mask

KW - Speech segregation

KW - Generalization

KW - Speech presence probability

U2 - 10.1109/IWAENC.2014.6953357

DO - 10.1109/IWAENC.2014.6953357

M3 - Article in proceedings

SN - 9781479968084

BT - Proceedings of IWAENC 2014

PB - IEEE

T2 - The International Workshop on Acoustic Signal Enhancement

Y2 - 8 September 2014 through 11 September 2014

ER -

Generalization of Supervised Learning for Binary Mask Estimation

Abstract

Conference

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this