Language is needed for any kind of communication, but the same facilitator of learning becomes a barrier when there are 2 people from different backgrounds and speaking in different languages. To fix this, Amazon has added a capability that offers real-time translations in multiple languages. Let’s find out what this new Amazon Alexa Live Translation feature is and how does it work!
How to use Amazon Alexa Live Translation on Echo devices?
If you have an Echo device like Echo Show set up in your living room, you can ask Alexa to start a translation session for a pair of languages. Once commenced, speak a few phrases or sentences in the supported languages. Alexa will quickly pick up which language you are speaking and translate each side of the conversation. See how to work with Alexa Live Translation.
For example, to use the issue Live Translation on Echo Show, issue a voice command like ‘Alexa, translate Spanish’. As soon as you hear the beep sound, begin speaking in the language you want to translate. Alexa will automatically detect the language being spoken and translate each side of the conversation. On Amazon Echo Show, in addition to hearing, you will see the translated language is displayed on the device’s screen.
When you’re done, end the translation session by saying ‘Alexa, stop.’
Which languages does Amazon’s Alexa Live Translation on Echo supports?
For now, the feature works with six language pairs on Echo devices with a locale set to English US. These include,
- Brazilian Portuguese
How does Amazon Live Translation Alexa work?
Alexa translate is optimized for conversational-speech translation. So it uses several of Amazon’s existing technology coupled with machine learning models, optimized for conversational-speech translation. These include,
- Amazon Translate.
- Alexa’s automatic-speech-recognition (ASR) system.
- Alexa’s text-to-speech system.
What is Amazon Translate and how does it work?
Amazon Translate is a service based on neural networks trained for language translation. It mainly works as an intermediary link between the source language and a target language. So, for the translation process to complete successfully, AT should receive a source text (text that you want to translate) as an input to generate the output text (the text translated into the target language).
Both, the Source Text and Output Text should be in UTF-8 format however, depending on the source and target languages, you can find more characters in the output text than in the input text.
The translation model that carries out the information processing has two components, namely,
- The Encoder – It reads a source sentence one word at a time and constructs a semantic representation that captures its meaning.
- The Decoder – generates translation for every single word in the target language by using the semantic representation.
Hope it helps!
Also read, – How to find a song by humming to Google app.