cedat logo cedat logo

What's new

Transcriptions 2.0 thanks to AI

Thursday January 13, 2022

Today, thanks to digitisation, it is becoming easier and easier to be helped by technology to complete jobs that were once considered as burdensome as transcribing a recording.

Not only that, AI will also be able to transcribe thoughts, let’s see how.

Audio files of different weights, sounds and languages. A tangle of information to be transcribed, and often to be translated, in a Word file. This tasks take time. For years, scholars of communication, scientists of the word, computer technicians, have put their hands and their heads on these issues, examining novelties and solutions that can help human beings and give them the necessary tools to allow them to free themselves from routine jobs.

Once upon a time, before speech to text or text to speech took hold, one would go to the PC, turn on the recorder, listen to the recorded video and report back step by step by tapping on the PC keys; a job that required hours and hours bending over the keyboards.

Today this is no longer the case because the tools are here. The transcription of an audio voice recording can be very useful in different contexts, just think of business, in which meetings require the drafting of a report, difficult to draw up when the meeting itself takes place, therefore it becomes necessary to make an audio recording which, subsequently, must be transcribed, that is, listened to and transcribed. Another example could be that of lectures at school or university, where the audio recording and subsequent transcription could be a useful support for the preparation of notes on which to study. The manual transcription operation is very long and burdensome and could take up to 5 or 6 times the duration time of the recorded material.

Thanks to technology, these hours and hours of work can be drastically reduced. There are specialized companies that have accelerated on this front, releasing solutions useful to humans that exploit the potential of artificial intelligence.

But today we can go even further. Thanks to researchers from the MIT Media Lab, a tool was created that would allow us not to lose our ‘silent conversations’ and to put them in writing, for example, on a PC or other digital devices. Are you cooking and would like to write down notes that came to your mind but you can’t? This tool transcribes these thoughts for you. Specifically, earpiece sensors read the signals that the wording sends from the brain to the areas the person would have activated if he had spoken aloud. Science fiction, you might say, may be, but think about what benefits it might have, for example for ALS sufferers.

Thanks to deep learning techniques supported by artificial intelligence, modern systems have significantly improved machine translation, but problems remain. For example, there are still areas, such as in social networks, which are unable to verify inappropriate content precisely because the artificial intelligence, in use, cannot decipher some foreign languages. This is a challenge that we want to launch and that we are sure will be met with results.