German Lombard conversation recordings

<p>These sound files contain conversations of three actors (one female and two male speakers) sitting in a cafeteria, simulating a daily life conversation of three students about six different topics in German. Lombard speech, where speaking in a noisy environment leads to an elevated amplitude and frequencies [1], was generated by providing the speakers with headphones with background noise during the recordings.</p> <p>These recordings can be used as speech material in hearing studies, for example. The recordings were used in [2].</p> <p>Files</p> <p>Lombard_conversations_script.pdf: Script of the conversations</p> <p>Wav-files (three speakers in three channels):</p> <p>bg_story_1.wav: Studying for exams</p> <p>bg_story_2.wav: Leisure activities and appointment at a restaurant</p> <p>bg_story_3.wav: Semester break plans and jobs</p> <p>bg_story_4.wav: Sports</p> <p>bg_story_5.wav: Pets</p> <p>bg_story_6.wav: Weather and appointment for a bike tour</p> <p>*.license: License abbreviation and author information, for use in the acoustic simulation tool TASCAR [3]</p> <p>two_channel_playback.m: Matlab/Octave script to play back the .wav files on two channels</p> <p>two_channel_playback.tsc: TASCAR [3] example scene playing back the .wav files on two channels</p> <p><br> &nbsp;</p> <p>Methods</p> <p>In preparation of the recordings, Sennheiser HDA200 headphones were calibrated with a Br&uuml;el &amp; Kj&aelig;r 4153 artificial ear and a 92 dB calibrator. A sound file containing a background noise recorded in a cafeteria was then calibrated to 75 dB. For the recordings, the background noise was played back by the Sennheiser HDA200 headphones which the speakers were wearing. Additionally, each speaker was equipped with a cardioid microphone and a script of the conversations. Then, the speakers spoke the conversations by improvising the text of the script, and the speech was recorded by the microphones. From the resulting recordings, distracting sounds such as crosstalk from the other speakers, page-flipping or verbal slips and errors were cut out and the three channels were combined and saved in sound files.</p> <p><br> &nbsp;</p> <p>References</p> <p>[1] Junqua, Jean‐Claude. &quot;The Lombard reflex and its role on human listeners and automatic speech recognizers.&quot; The Journal of the Acoustical Society of America 93.1 (1993): 510-524.</p> <p>[2] Hendrikse, MME, Eichler, T Grimm, G and Hohmann, V. (in preparation) Interaction of hearing aids with self-motion and the influence of hearing impairment</p> <p>[3] Grimm, Giso; Luberadzka, Joanna; Hohmann, Volker. A Toolbox for Rendering Virtual Acoustic Environments in the Context of Audiology. Acta Acustica united with Acustica, Volume 105, Number 3, May/June 2019, pp. 566-578(13), doi:10.3813/AAA.919337</p>