Dark Posted November 27, 2020 Posted November 27, 2020 Google has announced a new function in its recorder that through 'machine learning', or automatic learning, is capable of automatically marking the most important words of voice transcripts. The new function, called 'Smart Scrolling', is able to extract the keywords from each section of the transcripts and choose which parts of the text are the most informative. "Smart Scrolling offers additional text navigation capabilities that will further enhance the usefulness of the recorder, allowing users to quickly display sections of interest, even for long recordings," noted Google on its blog. To select keywords in the text, the model uses the ratio of the number of occurrences of a word compared to the conversational data set, but also takes into account how specific or broad the term is. Likewise, Google has indicated that the models used for this function are "light enough to run on the device without the need to load the transcript, thus preserving the privacy of the user." So far, the company has tested this feature on its Pixel 5 phones, but has not offered details on when it will be released for all devices.
Recommended Posts