Google's New AI Transcribe Feature - A Translating Wiz

  • Is the improved transcription feature the new replacement of the earlier Google Live Transcribe?
  • The latest audio-to-text translation service is out and about, but only for Android users for the time being. iPhone owners must wait a little longer.
  • Record the audio in one language and have it rendered in another language altogether!
  • Lengthy discussions can be easily transcribed into text now, without any trouble.

January marked the launch of the AI-Powered transcription feature of Google Translate on Android, and now it supports transcribed translations between any of the eight languages, including French, German, Portuguese, English, Thai, Hindi, Spanish, Russian.

This feature allows Google Translate users to transcribe translated audio into text...in a jiffy! Real-time multilingual transcription is a lot more complicated than it sounds! It’s very different from simply translating written text from one language to another or inputting single sentences of speech and rolling them into text in a different language.

In the past, Translate did not transcribe longer discussions, like those in a conference or lecture or maybe a story-telling session. However, this issue has been resolved and improvised through this recently launched AI-powered transcription feature. Although it does not work with audio files for now, as it needs to be live audio captured through a smartphone microphone, but Google suggests playing recorded audio through a speaker and capture it that way.

This feature is not available on the iOS Google Translate app as yet, but for the Android users, all that needs to be done is to install an update of the Google Translate app, then hit the “Transcribe” option, choose the source and target languages, tap the mic button and START!

It’s actually that simple!

The basic principle of mechanism is built on Google’s Live Transcribe Feature, which offers hearing-impaired users a real-time captioning service. Both Google Live Transcribe and Google Translate Transcribe features are powered by Google Cloud and its Tensor Processing Units (TPU), so transcription doesn't happen on device.

The Google Transcribe feature is based on a combination of Google's real-time translations that happen continuously with TPU hardware, which basically stacks machine translation on top of a real-time automatic speech-recognition system, thus allowing the system to generate a new translation for every update to a recognizable transcript.

The Live Transcribe feature follows the same technology that helps in automatic speech recognition, which also provides automated captions on YouTube videos and Google Slides presentations. This transcription feature constantly evaluates whole sentences as the audio goes on, adding punctuation, correcting certain word choices while gauging the sentence’s context, and improvising things like accents and regional dialects.


Through all these added benefits, users will get an accurate approximation of what is being said, which is a pretty amazing feat on its own!

Having said all that, the Google Translate Transcribe has been launched very recently, and despite its amazing and very promising initial results, there is always room for improvement, because honestly, from Google, the world expects nothing but perfection! And that is why we will keep hoping for high functioning, far improved underlying AI models with the passage of time.



Read next: Google's New 3D Object Recognition Process Might Soon Make Great AR Experiences Possible!

No comments:

Post a Comment