Grant: Discovery Grants Program - Individual
École de technologie supérieure
$25,000.00
Speech Analytic - From Speech to Emotion Recognition
EF 2014-15
Résumé
Advances in automatic speech recognition do not give us access to ever increasing amounts of unstructured audio files found on the Web in the form of videos, podcasts, TV shows, online courses, etc. Indeed, a number of unresolved issues stand between a verbatim transcription and automatic operation of the content: voice activity detection, speech recognition, speaker recognition, emotion detection, language detection, speaker turn segmentation, topic detection, phrase and paragraph segmentation. What is seen now as a multitude of individual steps is essential for the extraction of semantic content and subsequent treatments such as indexing, named entity detection and automatic summarization. The processing of such information is the aim of this proposal. The objective of this research project is to enable information extraction from highly unstructured audio data and continuously evolving at the pace of change of the day. In the short term, we will continue our works on speech recognition, speaker recognition, emotion detection and topic segmentation independently. We will also start a new topic on audio copy detection. Exploring many application domains will stimulate cross-fertilization of the models, exchange of data (for example, transcription from speech recognition can be used as text input for emotion detection and speaker verification which usually used only acoustic parameters) and interesting discussions among the research team. In the long term, we will like to explore new paradigm such as use of continuous data acquisition and human-guided machine learning.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
170388-2012
Name nom
Dumouchel, Pierre
Department département
Génie logiciel et des TI
Organization id
53
Institution établissement
École de technologie supérieure
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2014
Competition year année de concours
2012
Award amount
25000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1510
Committee name en
Electrical and Computer Engineering
Committee name fr
Génie électrique et informatique
Area of application code
800
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information and communication services
Area of application fr
Services d'information et de télécommunications
Research subject code
2801
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Natural language and speech understanding
Research subject fr
Langage naturel et reconnaissance de la parole
Partie
3
Nb partie
5
Application title
Speech Analytic - From Speech to Emotion Recognition
Keyword
NSERC
Application summary
Advances in automatic speech recognition do not give us access to ever increasing amounts of unstructured audio files found on the Web in the form of videos, podcasts, TV shows, online courses, etc. Indeed, a number of unresolved issues stand between a verbatim transcription and automatic operation of the content: voice activity detection, speech recognition, speaker recognition, emotion detection, language detection, speaker turn segmentation, topic detection, phrase and paragraph segmentation. What is seen now as a multitude of individual steps is essential for the extraction of semantic content and subsequent treatments such as indexing, named entity detection and automatic summarization. The processing of such information is the aim of this proposal. The objective of this research project is to enable information extraction from highly unstructured audio data and continuously evolving at the pace of change of the day. In the short term, we will continue our works on speech recognition, speaker recognition, emotion detection and topic segmentation independently. We will also start a new topic on audio copy detection. Exploring many application domains will stimulate cross-fertilization of the models, exchange of data (for example, transcription from speech recognition can be used as text input for emotion detection and speaker verification which usually used only acoustic parameters) and interesting discussions among the research team. In the long term, we will like to explore new paradigm such as use of continuous data acquisition and human-guided machine learning.