Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Engage Grants Program

Institut national de la recherche scientifique

$25,000.00

Room acoustics aware speech enhancement

EF 2013-14

Résumé
Recent projections from Global Industry Analysis have suggested that the global speech technology market willsurpass the $31 billion mark by 2017, mostly due to three applications: automatic speech recognition, speakerverification, and text-to-speech synthesis. Existing speech technologies, however, are extremely sensitive tovarying room acoustics, particularly room reverberation. As shown in a recent survey, speech recognitionperformance can go from near-perfect accuracy in clean conditions to almost 80% error rate in a reverberantroom. After speech enhancement, the error rate drops to about 30-40%. While this can be seen as a significantgain, the overall performance is still far below acceptable levels. One reason for such poor performance lies onlimitations of existing dereverberation algorithms, which commonly attempt to blindly estimate the inversefilter of the room impulse response (RIR). This is an ill-conditioned problem, as the RIR continuously changeswith factors such as acoustic source positioning, room temperature, and placement of room furnishings.Existing adaptive systems rely on suboptimal least-square error optimization paradigms, thus leading tosuboptimal results (43% word accuracy rate). To bridge this gap, incorporation of room properties, such as theso-called reverberation time and direct-to-reverberant energy ratio, into the optimization procedure is neededbefore acceptable speech recognition results are obtained in highly reverberant environments. This is the aim ofthe present project. More specifically, the main objective of the proposed research is to develop an adaptive""room-acoustics aware"" speech enhancement algorithm that is low in computational complexity. The algorithmwill build on the applicant's blind room acoustic characterization expertise and will explore the benefits ofsingle and multi-microphone solutions. The developed algorithm will enable our industry partner - UCIC - todevelop numerous applications, including voice-enabled Internet searches, home automation, and assistivedevices for elderly and/or for individuals with visual, hearing, or mobility impairments. The project also hasthe potential of impacting other areas, such as teleconferencing and hearing aids.
Programme
Engage Grants Program
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
452286-2013
Name nom
Falk, Tiago
Department département
Centre Énergie, Matériaux et Télécommunication - Montréal
Organization id
56
Institution établissement
Institut national de la recherche scientifique
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2013
Competition year année de concours
2013
Award amount
25000
Program id
EGP
Program name en
Engage Grants Program
Program name fr
Programme de subventions d'engagement partenariat
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
1552
Committee name en
Quebec Internal Decision Committee
Committee name fr
Comité de décision interne pour le Québec
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
2514
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Digital signal processing
Research subject fr
Traitement des signaux numériques
Partie
1
Nb partie
1
Application title
Room acoustics aware speech enhancement
Keyword
NSERC
Application summary
Recent projections from Global Industry Analysis have suggested that the global speech technology market willsurpass the $31 billion mark by 2017, mostly due to three applications: automatic speech recognition, speakerverification, and text-to-speech synthesis. Existing speech technologies, however, are extremely sensitive tovarying room acoustics, particularly room reverberation. As shown in a recent survey, speech recognitionperformance can go from near-perfect accuracy in clean conditions to almost 80% error rate in a reverberantroom. After speech enhancement, the error rate drops to about 30-40%. While this can be seen as a significantgain, the overall performance is still far below acceptable levels. One reason for such poor performance lies onlimitations of existing dereverberation algorithms, which commonly attempt to blindly estimate the inversefilter of the room impulse response (RIR). This is an ill-conditioned problem, as the RIR continuously changeswith factors such as acoustic source positioning, room temperature, and placement of room furnishings.Existing adaptive systems rely on suboptimal least-square error optimization paradigms, thus leading tosuboptimal results (43% word accuracy rate). To bridge this gap, incorporation of room properties, such as theso-called reverberation time and direct-to-reverberant energy ratio, into the optimization procedure is neededbefore acceptable speech recognition results are obtained in highly reverberant environments. This is the aim ofthe present project. More specifically, the main objective of the proposed research is to develop an adaptive""room-acoustics aware"" speech enhancement algorithm that is low in computational complexity. The algorithmwill build on the applicant's blind room acoustic characterization expertise and will explore the benefits ofsingle and multi-microphone solutions. The developed algorithm will enable our industry partner - UCIC - todevelop numerous applications, including voice-enabled Internet searches, home automation, and assistivedevices for elderly and/or for individuals with visual, hearing, or mobility impairments. The project also hasthe potential of impacting other areas, such as teleconferencing and hearing aids.
Institut national de la recherche scientifique - $25,… | Canada Spends