Localization in Static and Dynamic Hearing Scenarios: Utilization of Machine Learning and Binaural Auditory Model

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F17%3A00312007" target="_blank" >RIV/68407700:21230/17:00312007 - isvavai.cz</a>

Výsledek na webu

<a href="http://radio.feld.cvut.cz/conf/poster/proceedings/Poster_2017/Section_NS/NS_061_Koshkina.pdf" target="_blank" >http://radio.feld.cvut.cz/conf/poster/proceedings/Poster_2017/Section_NS/NS_061_Koshkina.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Localization in Static and Dynamic Hearing Scenarios: Utilization of Machine Learning and Binaural Auditory Model

Popis výsledku v původním jazyce

Hearing with both ears, in other words binaural hearing, allows human to localize sound sources in a space. Models of binaural hearing often simulate functions of lateral and medial superior olives (LSO and MSO), but their outputs cannot be in most cases directly mapped to certain azimuths in space. In this paper we present an azimuth classification algorithm, which utilizes both binaural models (LSO and MSO) for preprocessing of a sound signal. From their outputs features are extracted for machine learning algorithms: K-Nearest Neighbors and Artificial Neural Network. The algorithm is trained and tested on speech samples from ITU - T Rec. P501 and NOIZEUS corpora. The success of the K-NN and ANN classifiers is discussed. Both machine learning algorithms give a similar classification error in static and dynamic hearing scenarios. The error s comparable to human psychoacoustical data.

Název v anglickém jazyce

Localization in Static and Dynamic Hearing Scenarios: Utilization of Machine Learning and Binaural Auditory Model

Popis výsledku anglicky

Hearing with both ears, in other words binaural hearing, allows human to localize sound sources in a space. Models of binaural hearing often simulate functions of lateral and medial superior olives (LSO and MSO), but their outputs cannot be in most cases directly mapped to certain azimuths in space. In this paper we present an azimuth classification algorithm, which utilizes both binaural models (LSO and MSO) for preprocessing of a sound signal. From their outputs features are extracted for machine learning algorithms: K-Nearest Neighbors and Artificial Neural Network. The algorithm is trained and tested on speech samples from ITU - T Rec. P501 and NOIZEUS corpora. The success of the K-NN and ANN classifiers is discussed. Both machine learning algorithms give a similar classification error in static and dynamic hearing scenarios. The error s comparable to human psychoacoustical data.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20201 - Electrical and electronic engineering

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2017

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the International Student Scientific Conference Poster – 21/2017

ISBN

978-80-01-06153-4

ISSN

—

e-ISSN

—

Počet stran výsledku

5

Strana od-do

1-5

Název nakladatele

Czech Technical University in Prague

Místo vydání

Praha

Místo konání akce

Praha

Datum konání akce

23. 5. 2017

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—