Speaker Diarization of Broadcast Streams using Two-stage Clustering based on I-vectors and Cosine Distance Scoring

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F46747885%3A24220%2F12%3A%230002004" target="_blank" >RIV/46747885:24220/12:#0002004 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Speaker Diarization of Broadcast Streams using Two-stage Clustering based on I-vectors and Cosine Distance Scoring

Popis výsledku v původním jazyce

In this paper we present our system for speaker diarization of broadcast news based on recent advances in the speaker recognition field. In the system, speaker segments determined by the speaker changepoint detector are represented by i-vectors and similarity of segments? speakers evaluated using cosine distance scoring. Linear discriminant analysis is employed to cope with intra-speaker variability. The experiments were carried out using the COST278 multilingual broadcast news database. We demonstrateimprovement of the performance over the baseline system based on the Bayesian Information Criterion (BIC) and highlight significant impact of cepstral mean normalization. Finally, two-stage clustering employing BIC-based clustering to pre-cluster segments in the first stage is examined and showed to yield further performance improvement. The best performing configuration of our system achieved 52.4 % relative improvement of the speaker error rate over the baseline.

Název v anglickém jazyce

Speaker Diarization of Broadcast Streams using Two-stage Clustering based on I-vectors and Cosine Distance Scoring

Popis výsledku anglicky

In this paper we present our system for speaker diarization of broadcast news based on recent advances in the speaker recognition field. In the system, speaker segments determined by the speaker changepoint detector are represented by i-vectors and similarity of segments? speakers evaluated using cosine distance scoring. Linear discriminant analysis is employed to cope with intra-speaker variability. The experiments were carried out using the COST278 multilingual broadcast news database. We demonstrateimprovement of the performance over the baseline system based on the Bayesian Information Criterion (BIC) and highlight significant impact of cepstral mean normalization. Finally, two-stage clustering employing BIC-based clustering to pre-cluster segments in the first stage is examined and showed to yield further performance improvement. The best performing configuration of our system achieved 52.4 % relative improvement of the speaker error rate over the baseline.

Druh

D - Stať ve sborníku

CEP obor

JC - Počítačový hardware a software

OECD FORD obor

—

Projekt

<a href="/cs/project/TA01011204" target="_blank" >TA01011204: Živé archivy</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2012

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proc. of International Conference on Acoustics, Speech, and Signal Processing - ICASSP 2012

ISBN

978-1-4673-0046-9

ISSN

—

e-ISSN

—

Počet stran výsledku

4

Strana od-do

4193-4196

Název nakladatele

—

Místo vydání

Japonsko

Místo konání akce

Tokyo, Japonsko

Datum konání akce

1. 1. 2012

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000312381404066