Training with time aligned transcription

[Erik Sigcha]

Hi
I want to know if it is possible to train an acoustic model with time aligned transcriptions like this:

0.000000 59.491978 .sil
59.491978 259.358276 cua_7l
259.358276 323.529419 es
323.529419 447.860962 la
447.860962 953.208557 diferencia
953.208557 1100.267334 de
1100.267334 1267.379639 este
1267.379639 1961.898438 gobierno
1961.898438 2019.365079 .sil

Thanks

[Nickolay V. Shmyrev]

Yes, you can train, but alignment will be simply dropped. It is not needed anyway.