Amazon Adds Live Translation to Alexa Features
Amazon has introduced the new Live Translation feature to Alexa, enabling real-time translations between certain languages in both voice and text form. The feature uses the same AI models as Alexa’s bilingual understanding to recognize which side of several pairs of languages is being spoken and translating to the other. Right now, the translations are limited to English and with French, Spanish, Hindi, German, Italian, or Brazilian Portuguese.
Translate Talk
Live Translate available on any Echo device by asking Alexa in English to translate German or French, or any of the other languages. When the voice assistant beeps, the user can speak either language naturally and Alexa will subsequently repeat back what was said in the other language. The voice assistant will identify automatically which of the two languages is being spoken during the conversation, and will display the words as well as speak them if the Echo device has a screen.
“During a translation session, Alexa runs two ASR models in parallel, along with a separate model for language identification,” Alexa AI senior manager for translation Shirin Saleem Alexa speech applied-science manager Roland Maas explained in a blog post. “Input speech passes to both ASR models at once. Based on the language ID model’s classification result, however, only one ASR model’s output is sent to the translation engine.”
Lingua Alexa
In some ways, the new feature just extends the multilingual mode that Alexa added in 2018 and recently expanded to a bunch of new languages. With the update, Alexa can identify and respond to English or one of several additional languages including German, Italian, Japanese, Hindi, both Canadian and standard French, as well as standard Spanish and the dialect spoken in the U.S. Multilingual mode was built for serving bilingual homes where people switch between languages all the time, but the utility of the dual-language model operation extends well beyond that, as Live Translate makes clear.
Amazon is pitching Live Translate as an ideal way to improve communication between people who don’t speak the same language well, as a practice tool for learning a new tongue, or as a way to connect hotel guests and staff if Alexa for Hospitality devices are in use. Live Translate could also be useful when driving, a variation of the multilingual mode incorporated into the newest version of the Alexa Auto SDK. Translating languages is a potentially very powerful application of voice AI, and Alexa is facing plenty of competition from Google, Facebook, and others. The next step for Alexa will be understanding issues like language formality, context, and colloquialisms, all of which are in the blueprint for future updates to the service.
“To improve the fluency of the translation and its robustness to spoken-language input, we are also working on adapting the neural-machine-translation engine to conversational-speech data and generating translations that incorporate relevant context, such as tone of voice or formal versus informal translations,” the blog’s authors concluded. “Finally, we are continuously working on improving the quality of the overall translations and of colloquial and idiomatic expressions in particular.”
Follow @voicebotai Follow @erichschwartz
Facebook Builds Speech Recognition Engine Combining 51 Languages in One Model
