Logo des Repositoriums
 

Prosodic addressee-detection: ensuring privacy in always-on spoken dialog systems

Kurzanzeige
dc.contributor.authorBaumann, Timo
dc.contributor.authorSiegert, Ingo
dc.contributor.editorAlt, Florian
dc.contributor.editorSchneegass, Stefan
dc.contributor.editorHornecker, Eva
dc.date.accessioned2020-09-16T07:52:29Z
dc.date.available2020-09-16T07:52:29Z
dc.date.issued2020
dc.description.abstractWe analyze the addressee detection task for complexityidentical dialog for both human conversation and devicedirected speech. Our recurrent neural model performs at least as good as humans, who have problems with this task, even native speakers, who profit from the relevant linguistic skills. We perform ablation experiments on the features used by our model and show that fundamental frequency variation is the single most relevant feature class. Therefore, we conclude that future systems can detect whether they are addressed based only on speech prosody which does not (or only to a very limited extent) reveal the content of conversations not intended for the system.en
dc.description.urihttps://dl.acm.org/doi/10.1145/3404983.3410021en
dc.identifier.doi10.1145/3404983.3410021
dc.identifier.urihttps://dl.gi.de/handle/20.500.12116/34263
dc.language.isoen
dc.publisherACM
dc.relation.ispartofMensch und Computer 2020 - Tagungsband
dc.relation.ispartofseriesMensch und Computer
dc.subjectfundamental frequency variation
dc.subjectaddressee detection
dc.subjectrecurrent neural network
dc.subjectcomplexity-identical human-computer interaction
dc.subjectcomputational paralinguistics
dc.titleProsodic addressee-detection: ensuring privacy in always-on spoken dialog systemsen
dc.typeText/Conference Paper
gi.citation.publisherPlaceNew York
gi.citation.startPage195–198
gi.conference.date6.-9. September 2020
gi.conference.locationMagdeburg
gi.conference.sessiontitleMCI: Short Paper (Poster)
gi.document.qualitydigidoc

Dateien

Sammlungen

Tagungsband MuC 2020