Neural nets translate speech straight to text

• Autores: Matt Reynolds
• Localización: New scientist, ISSN 0262-4079, Nº. 3120, 2017, pág. 9
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Google's latest take on machine translation could make it easier for people to converse in different languages, by translating speech in one language directly into text in another. Researchers at Google Brain, the tech giant's deep learning research arm, have turned to neural networks. By skipping transcription, the approach could potentially allow for more accurate and quicker translations. The team trained its system on hundreds of hours of Spanish audio with corresponding English text. It used several layers of neural networks--computer systems loosely modeled on the human brain--to match sections of the spoken Spanish with the written translation. To do this, it analyzed the waveform of the Spanish audio to learn which parts seemed to correspond with which chunks of written English. When it was then asked to translate, each neural layer used this knowledge to manipulate the audio waveform until it was turned Into the corresponding section of written English