Speech recognition is available through Tiro, which developed a speech synthesiser for CDHA. Users can upload audio files or provide a link to a webpage containing an audio file, which will then be analysed and transcribed.
Tiro offers several speech recognition models, but users can select Icelandic (ÍSMÚS), which is specifically designed for older recordings with more archaic language or poor recording quality.
The speech recognition tool is free to use, but registration is required.
The tool was developed for use with audio recordings related to Icelandic cultural heritage (ÍSMÚS). On the ÍSMÚS website, users can now listen to recordings and read automatically transcribed text simultaneously.
It is important to exercise caution when reading the transcriptions, as the quality of the transcribed text may vary.
ÍSMÚS is managed by the Árni Magnússon Institute for Icelandic Studies and the National and University Library of Iceland.
The training data used to develop the speech recognition model for older Icelandic is stored and accessible through CLARIN Iceland. It may be used under the licensing terms published by CLARIN-IS.
The project was supported by the Infrastructure Fund.
