search query: @keyword speech / total: 10
reference: 1 / 10
« previous | next »
Author: | MacĂas Ojeda, Antonio |
Title: | Speaker Diarization |
Publication type: | Master's thesis |
Publication year: | 2014 |
Pages: | ix + 30 Language: eng |
Department/School: | Perustieteiden korkeakoulu |
Main subject: | Machine Learning and Data Mining (SCI3015) |
Supervisor: | Kurimo, Mikko |
Instructor: | |
Electronic version URL: | http://urn.fi/URN:NBN:fi:aalto-201412033122 |
OEVS: | Electronic archive copy is available via Aalto Thesis Database.
Instructions Reading digital theses in the closed network of the Aalto University Harald Herlin Learning CentreIn the closed network of Learning Centre you can read digital and digitized theses not available in the open network. The Learning Centre contact details and opening hours: https://learningcentre.aalto.fi/en/harald-herlin-learning-centre/ You can read theses on the Learning Centre customer computers, which are available on all floors.
Logging on to the customer computers
Opening a thesis
Reading the thesis
Printing the thesis
|
Location: | P1 Ark Aalto 2461 | Archive |
Keywords: | speech speaker recognition diarization |
Abstract (eng): | In this thesis we document the development of a system to perform Speaker Diarization, that is, automatically trying to identify who spoke when in a conversation or any other piece of speech with several speakers. The intended usage is to be able to provide this functionality for broadcast news, with data provided by the Finnish broadcasting company YLE under the Next Media programme, financed by TEKES, the Finnish Funding Agency for Technology and Innovation. Another goal is to produce a system compatible with existing Aalto University speech recognition software, in order to open the door to future improvements and research. The produced system, a newly implementation of established methods, with the parameters we determined were the best for our use case, obtains a performance that is very close to current stat-of-the-art systems, while still being compatible with the existing speech recognition software of the Aalto University and having a reasonable speed performance. Further improvements to the system are being made as we speech, opening the door to more research options. |
ED: | 2014-12-21 |
INSSI record number: 50207
+ add basket
« previous | next »
INSSI