Section: New Software and Platforms

Data acquisition

Participants : Vincent Colotte, Slim Ouni, Yves Laprie.

The nature of our research makes us highly concerned by acquisition and processing of speech data. Besides acquisition of speech audio signals, we are concerned with the acquisition of articulatory data, mainly ElectroMagnetographic Articulography (EMA) data using an articulograph and Magnetic Resonance Imaging (MRI) data. EMA captures articulatory movements in three dimensions (3D) with a high temporal resolution by tracking tiny sensors attached to speech articulators such as the tongue, teeth, and lips. MRI is a non-invasive, hazard-free medical imaging technique allowing for high-resolution scans of the vocal tract.

JCorpusRecorder

JCorpusRecorder is a software for the recording of audio corpora. It provides an easy tool to record with a microphone. The audio input gain is controlled during the recording. From a list of sentences, the output is a set of wav files automatically renamed according to textual information given in input (nationality, speaker language, gender...). An easy to use tagging allows for displaying a textual/visual/audio context of the sentence to pronounce. This software is suitable for recording sentences with information to guide the speaker. The sentences can be presented randomly. The software is developed in Java. It is currently used for the recording of sentences in several projects.

EMA acquisition platform

Since the purchase of the articulograph AG500 in 2007, we have built a strong experience with respect to the acquisition technique and we have developed an acquisition protocol (sterilization, calibration, etc.). The platform has been improved by acquiring the latest articulograph AG501 funded by the EQUIPEX ORTOLANG project. The AG501 allows tracking the movement of 24 sensors at reasonable high frequency (250Hz) to very high frequency (1250Hz). In addition, we have developed a powerful tool, VisArtico, to visualize articulatory data acquired using an articulograph.

MRI acquisition platform

Magnetic Resonance Imaging (MRI) takes an increasing place in the investigation of speech production because it provides a complete geometrical information of the vocal tract. We thus initiated a cooperation with the IADI laboratory (Imagerie Adaptive Diagnostique et Interventionnelle) at Nancy Hospital, which studies in particular magnetic resonance imaging. This year, we acquired static MRI data for two speakers (approximately 90 blocked articulations corresponding to vowels and consonants followed by a vowel) and we carried out preliminary experiments intended to acquire dynamic data.