Menlo Park, US – Google (NASDAQ: GOOGL) finally decided to improve its voice transcription feature in Google Voice. The voicemail and phone management system by Google has created quite a lot bloopers in the past, thanks to Google’s voice transcription. The search giant mentioned in one of its official blogs that the updated voice transcription of Google Voice is making 49% less errors. For this update, Google has used ‘long short-term memory deep recurrent neural network’. Google says that users don’t have to do anything extra to use this update. The existing users can continue with the Google Voice service, as it will be updated automatically.
Google Voice: Bloopers and updates
Yesterday, on July 23rd, Zander Danko, Software Engineer at Google, posted to his blog that Google Voice will now have much improved voice transcription. In the past, Google has received several complaints that the message transcribed by Google Voice does not match exactly with the original voicemail. In fact, people have found quite funny messages left by the voice transcription. If you want to read the hilarious bloopers that Google Voice has done in the past, read them here.
In order to tackle the problem and reduce the errors in the transcribed messages, Google asked users to submit their voice mails and transcribed messages. Due to this request, several Google Voice users participated and this data was used for research and system improvements. The updated voice transcription service makes use of ‘long short-term memory deep recurrent neural network’. This updated service will be a part of Google Voice as well as Google’s latest fast and easy Wi-Fi experience, called as ‘Project-Fi’.
“Google Voice gives you one number for all of your phones — a phone number that is tied to you, not to a device or a location. Make and pick up calls, send and get texts, and read and listen to voicemails on your phone, tablet, or computer.”
Those who have not yet registered on Google Voice can register themselves and also see all the features of Google Voice on this link.