Improvements

• New Finnish models for Enhanced Operating Point

  • 20% relative accuracy improvement

  • Improved accuracy for transcription of medical terminology

• New models for Standard Operating Point, with improvements in discriminating between speech and non-speech audio

  • Bashkir (ba), Bashkir (eu), Belarusian (be), Bengali (bn), Bulgarian (bg), Catalan (ca), Croatian (hr), Czech(cs), Esperanto (eo), Estonian (et), Finnish (fi), Galician (gl), Greek (el), Hebrew (he), Interlingua (ia), Irish (ga), Latvian (lv), Lithuanian (lt), Malay (ms), Malay & English bilingual (en_ms), Maltese (mt), Mandarin & English bilingual (cmn_en), Marathi (mr), Mongolian (mn), Persian (fa), Polish (pl), Portuguese (pt), Romanian (ro), Russian (ru), Slovakian (sk), Slovenian (sl), Swahili (sw), Swedish (sv), Tagalog (tl), Tamil (ta), Tamil & English bilingual (en_ta), Thai (th), Turkish (tr), Ukrainian (uk), Urdu (ur), Uyghur (ug), Vietnamese (vi), Welsh (cy)