Spoken language identification over fixed-length windows of an audio track, using the MMS language identification models. Consecutive windows with the same language can be merged into one TimeFrame.
@keighrim