Reference

Jochim, M., Draxler, C. (2016). Fully Automated Accent Correction for Computer-Assisted Speech Rhythm Training: A Pilot Study. In Tagungsband Der 12. Tagung Phonetik Und Phonologie Im Deutschsprachigen Raum (pp. 74-77).

BibTeX

@inproceedings{jochimFullyAutomatedAccent2016,
  title = {Fully {{Automated Accent Correction}} for {{Computer-Assisted Speech Rhythm Training}}: {{A Pilot Study}}},
  booktitle = {Tagungsband Der 12. {{Tagung Phonetik}} Und {{Phonologie}} Im Deutschsprachigen {{Raum}}},
  author = {Jochim, Markus and Draxler, Christoph},
  year = {2016},
  pages = {74--77},
  address = {M{\"u}nchen, Deutschland},
  abstract = {In this project we have implemented a feedback method for CAPT (computer-assisted pronunciation training), where the learner's syllable durations are corrected in a fully automated way. The correction is based on the PSOLA algorithm [1] and uses a reference speaker's recording to determine the desired syllable durations for a given utterance. Learners get to hear a pre-defined sentence in their target language and are expected to imitate it as closely as possible. Their recording is then manipulated to match the reference speaker's syllable durations and the manipulation is presented as auditory, corrective feedback. Exploiting this feedback, learners are expected to try again and improve their pronunciation in terms of speech rhythm. A pilot study with six learners of German with various L1 backgrounds was conducted to assess whether learners can successfully exploit the feedback to improve their pronunciation. The results suggest that learners can improve their pronunciation significantly using this method. They also suggest that the method works equally well or better than repeatedly listening to the reference speaker. Index Terms: computer-assisted}
}