Google demos real-time speech to text translation


Nicole Kobie

29 Jan, 2020

Google has demonstrated a real time translation and transcription tool powered by AI, that will take lectures and other long-form voice in one language and output it in another. 

Part of Google Translate, the tool will let a smartphone act as interpreter, listening to speech via the microphone and transcribing translated text in real time. So far, the system only supports a few languages, including English, French, German and Spanish. The demo showed English being translated to Spanish. 

“With this, your Android mobile phone will effectively turn into an almost real time translator device for long-form speech,” Google said at the demonstration according to reports, adding it could “unlock continuous speech translations in this world at scale” in the longer term.

However, the transcription and translation won’t happen on your device. Instead, the audio recording will be uploaded to Google’s servers – so you’ll need a decent Wi-Fi connection or solid data package, as well as a willingness to share the audio with Google, to use this tool. 

Google didn’t say when the feature would arrive. 

The Google Translate tool was unveiled as part of a showcase of Google’s AI projects at its San Francisco office, which also included a neural network used to track and monitor whales using underwater microphones, which is now being used in Canada.  

“With this information, marine mammal managers can monitor and treat whales that are injured, sick or distressed,” said Julie Cattiau, product manager for Google AI, in a blog post. “In case of an oil spill, the detection system can allow experts to locate the animals and use specialised equipment to alter the direction of travel of the orcas to prevent exposure.” 

That machine-learning system was trained on 1,800 hours of underwater audio recordings that were labelled and supplied by Fisheries and Oceans Canada. 

The AI demonstrations come a week after Google CEO Sundar Pichai wrote a column urging sensible regulation of artificial intelligence to avoid misuse of the technology.