Can you hear me now? Reducing the Lombard effect in a driving car using an In-Car Communication system

Publication: Research - peer-reviewArticle in proceedings

Documents

  • 15

    Final published version, 1 MB, PDF-document

DOI

  • Rabea Landgraf
  • Johannes Kohler-Kaeß
  • Christian Lüke
  • Oliver Niebuhr
  • Gerhard Schmidt

View graph of relations

This study aimed to evaluate an In-Car Communication system (ICC), which was developed to improve the communication between passengers inside a driving car. The evaluation was conducted by assessing parameters involved in the Lombard effect, i.e. modifications of speech production in the presence of loud noises. Speech recordings were made inside a stationary car at Kiel University using an acoustic and visual ambiance simulation to imitate real driving situations. In this way, background noises of different driving speeds can be removed from the signals after the recordings, thus allowing undisturbed analyses of acoustic parameters. Recordings were done at noise conditions of silence, 50 km/h and 130 km/h, with and without the use of ICC. 16 subjects participated in the production experiment. Analyses showed that - both with and without ICC - fundamental frequency and intensity increased at higher noise levels, thus confirming the Lombard effect. But, this phenomenon was reduced by the use of ICC, and both pitch and intensity decreased. Furthermore, the reduction of the Lombard effect due to ICC was greater in the back seat compared with the front seat.
Original languageEnglish
Title of host publicationProc. 8th International Conference of Speech Prosody
EditorsJon Barnes, Alejna Brugos, Stefanie Shattuck-Hufnagel, Nanette Veilleux
PublisherInternational Speech Communication Association (ISCA)
Publication date1 Jun 2016
Pages479-483
DOIs
StatePublished - 1 Jun 2016
Event8th Speech Prosody Conference - Boston, United States

Conference

Conference8th Speech Prosody Conference
LandUnited States
ByBoston
Periode31/05/201603/06/2016
SeriesSpeech Prosody
ISSN2333-2042

Download statistics

No data available