Cross-Lingual Adaptation of Broadcast Transcription System to Polish Language Using Public Data Sources

We present methods and procedures designed for cost-efficient adaptation of an existing speech recognition system to Polish. The system (originally built for Czech language) is adapted using common texts and speech recordings accessible from Polish web-pages. The most critical part, an acoustic model (AM) for Polish, is built in several steps, which include: (a) an initial bootstrapping phase that utilizes existing Czech AM, (b) a lightly-supervised iterative scheme for automatic collection and annotation of Polish speech data, and finally (c) acquisition of a large amount of broadcast data in an unsupervised way. The developed system has been evaluated in the task of automatic content monitoring of major Polish TV and Radio stations. Its transcription accuracy (measured on a set of 4 complete TV news shows with total duration of 105 min) is 79,2%. For clean studio speech, its accuracy gets over 92%.

[1]  Hermann Ney,et al.  Cross-language bootstrapping for unsupervised acoustic model training: rapid development of a Polish speech recognition system , 2009, INTERSPEECH.

[2]  Danijel Koržinek,et al.  Grammar Based Automatic Speech Recognition System for the Polish Language , 2007 .

[3]  Jan Nouza,et al.  Investigation of deep neural networks for robust recognition of nonlinearly distorted speech , 2014, INTERSPEECH.

[4]  Jan Silovský,et al.  Speech-to-text technology to transcribe and disclose 100, 000+ hours of bilingual documents from historical Czech and Czechoslovak radio archive , 2014, INTERSPEECH.

[5]  Mariusz Ziólko,et al.  Automatic Speech Recognition System Dedicated for Polish , 2011, INTERSPEECH.

[6]  K. Marasek.  Large vocabulary continuous speech recognition system for Polish , 2003 .

[7]  Robert Cecko,et al.  Development of Large Vocabulary Continuous Speech Recognition for Polish , 2012 .

[8]  Jan Nouza,et al.  Cost-Efficient Development of Acoustic Models for Speech Recognition of Related Languages , 2013 .

[9]  Jan Nouza,et al.  Using TTS for Fast Prototyping of Cross-Lingual ASR Applications , 2010, COST 2102 Conference.

[10]  Ngoc Thang Vu,et al.  Rapid bootstrapping of five eastern european languages using the rapid language adaptation toolkit , 2010, INTERSPEECH.

[11]  Tanja Schultz,et al.  Globalphone: a multilingual speech and text database developed at karlsruhe university , 2002, INTERSPEECH.

[12]  Leslaw Pawlaczyk,et al.  Skrybot - A System for Automatic Speech Recognition of Polish Language , 2009, ICMMI.

[13]  Grazyna Demenko,et al.  JURISDIC: Polish Speech Database for Taking Dictation of Legal Texts , 2008, LREC.

[14]  Ngoc Thang Vu,et al.  Multilingual a-stabil: A new confidence score for multilingual unsupervised training , 2010, 2010 IEEE Spoken Language Technology Workshop.

[15]  Jan Nouza,et al.  ASR for South Slavic Languages Developed in Almost Automated Way , 2016, INTERSPEECH.