Скачать презентацию Automated Translation Text to speech Скачать презентацию Automated Translation Text to speech

dab16cb312a69080f53c052781e95e0c.ppt

  • Количество слайдов: 14

Automated Translation Automated Translation

 • Text to speech • Speech recognition • Automated translations • Text to speech • Speech recognition • Automated translations

 • Babelfish: http: //babelfish. altavista. com/tr – Based on a system by Systran • Babelfish: http: //babelfish. altavista. com/tr – Based on a system by Systran – How accurate are translations? Where do they fail? Where do they succeed? • From English to Dutch and back again: Is the translation to be how accurate? Do them fail to where? Do them succeed to where?

Another example: • Babelfish: http: //babelfish. altavista. com/tr – I will be traveling to Another example: • Babelfish: http: //babelfish. altavista. com/tr – I will be traveling to the Netherlands in June with my two nephews and my 83 year old mother. English to Dutch and back again: – I will travel to the Netherlands in June with my two cousins and my 83 one person whose birthday it is mother. One more time: I to the Netherlands in June with my two cousins and my 83 one person travels of whom anniversary it is mother.

Problems • Word order • Preserving meaning • Idioms Problems • Word order • Preserving meaning • Idioms

 • Some data on the quality of three types of translations (Pérez-Quiñones et • Some data on the quality of three types of translations (Pérez-Quiñones et al. 2005) – Based on: Structure, Vocabulary (spelling, meaning, cognates), style (consistency, punctuation) and message

 • Some data (Pérez-Quiñones et al. 2005) • Some data (Pérez-Quiñones et al. 2005)

 • Some data (Pérez-Quiñones et al. 2005) • Some data (Pérez-Quiñones et al. 2005)

Applied Translations • What are some problems with using this technology for mainstream, commercial Applied Translations • What are some problems with using this technology for mainstream, commercial use? • Where can it be used?

The Future • Creating rigorous rules-based systems based on thousands of hours of research The Future • Creating rigorous rules-based systems based on thousands of hours of research and coding

The Future • Creating rigorous rules-based systems based on thousands of hours of research The Future • Creating rigorous rules-based systems based on thousands of hours of research and coding • Or let’s just fake it

Google • Compares texts that have been translated by humans to build a statistical Google • Compares texts that have been translated by humans to build a statistical model for translations • Asks for both user feedback and users to submit text they have already translated – Getting people to work for free! • More information: http: //www. google. com/help/faq_translation. html

 • Google: http: //www. google. com/help/faq_translation. html – I will be traveling to • Google: http: //www. google. com/help/faq_translation. html – I will be traveling to the Netherlands in June with my two nephews and my 83 year old mother. – From English to Dutch and back again: – I will travel to the Netherlands in June with my two nephews and my 83 year old mother.

Try it yourself! • Babelfish: http: //babelfish. altavista. com/tr • Google: http: //www. google. Try it yourself! • Babelfish: http: //babelfish. altavista. com/tr • Google: http: //www. google. com/help/faq_translation. html#whatis