[last update: 2024-02 / next update: 2025-02]
Autophon – Automatic phonetic annotation
of the Nordic languages
outputs
LINK | Young, N.J. (2023). DanFA 3.0 – Forced Alignment of Danish, version 3.0. www.autophon.se.
LINK | Young, N.J. & McGarrah, M. (2021). FAVE-Align for Nordic languages – Rapidly constructing a high-quality prototype for Swedish. Nordic Journal of Linguistics, 1–27.
LINK | Young, N.J. (2020). NoFA 1.0 – forced alignment for bokmål, version 1.0. Språkbanken, National Library of Norway. https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-59
LINK | Young, N.J. & McGarrah, M. (2017). LG-FAVE, version 1.0. https://github.com/mcgarrah/LG-FAVE
Funding
about
ongoing
Funding body: The Swedish Academy, University of Oslo, European Union's Marie Sklodowska-Curie grant agreement No 892963 (and self-financed)
Autophon is a free web app I built (www.autophon.se) that hosts forced alignment models for Danish, Norwegian bokmål, Swedish, and UK English. The app uses the Montreal Forced Aligner 1.0 together with some other tools to output time-stamped phonetic boundaries in Praat from the following user inputs: (1) sound files and (2) matching orthographic transcriptions.
The current Danish and Swedish models can be used through the web app, but they were privately financed and therefore not available for download. The Norwegian model, however, was publicly financed and is held by the Norwegian National Library here.
A new Swedish model will soon be available both for download and for use within the web app. The old model was solely trained on speech material from men, so I applied for and was granted funds by the Swedish Academy to retrain it with the speech of women. A new Norwegian model will also soon be available both for download and for use within the web app, funded by my ERC Marie-Sklodowska Curie Grant No 892963 and a grant from the Department of Linguistics and Scandinavian Studies at The University of Oslo.