Aalto-ASR – Aalto University Automatic Speech Recognition System


Aalto-ASR includes two main functionalities: automatic speech recognition and automatic forced alignment. The speech recognizer allows you to obtain a transcript of an audio file that contains Finnish speech. The result of the recognition process is returned either as an annotation file (such as Praat TextGrid) or as a plain text file. In case a transcript of the audio file already exists, the words included in the text can be automatically aligned with the corresponding parts of the audio signal. The automatic forced aligner also works in several other languages in addition to Finnish.

Aalto University Automatic Speech Recognition System, version 2.1
Metadata, license and citation instructions
Instructions (in Finnish only; English instructions forthcoming!)

Recent updates

The new, upgraded version 2.1 was installed in Kielipankki in September 2021. The tools can be used in the Puhti environment. If required, Aalto-ASR can now also be installed on other environments as a Docker container.

Some features of version 1 were previously available via the Mylly service as well. However, the old Aalto-ASR version was removed in 2020 and the new version has not yet been plugged in to Mylly, but this may happen at a later stage.


Persistent identifier of this resource group page: http://urn.fi/urn:nbn:fi:lb-2021082325