Autophon (beta) is a free prototype web app that uses
online forced alignment to convert an audio file and corresponding
transcript into a time-aligned phonetic annotation that is
readable in
Praat.
Forced alignment is technology that uses neural networks to
determine, for each phonetic segment of the transcript, the time
interval in the audio file that contains the spoken segment. The
backend is built on
The Montreal Forced Aligner, and the language-specific models have been predominantly
trained on naturally-occurring spontaneous speech.
A language-specific write-up with metrics can be accessed by
clicking on the relevant language in the list.
Danish (DanFA 3.0)
Danish (DanFA 4.0 ⚠️ dev mode)
Norwegian - Bokmål
Swedish - Sweden
English - United Kingdom
English - North America
French - France
French - Canada
German - Germany
Elfdalian
Faroese
Finnish
Greenlandic
Icelandic
Sami
﹡A liberal deployment of the term "Nordic".
If you already have an account, sign in here.
Dont have an account? Sign up here
Enter your email address, we'll email a link to change your password.
Dont have an account? Sign up here