Tech giant Google recently announced that it is upgrading tits Voice Search service for the Android and iOS platform. The upgrade will make it easier for the app to pick up user commands even if they are in a crowded or noisy place.
Google made the announcement on its official blog on Sept. 24. The tech giant posted, "Today, we're happy to announce we built even better neural network acoustic models using Connectionist Temporal Classification (CTC) and sequence discriminative training techniques."
Google added these upgrades are an extension to the Recurrent Neural Networks (RNN) which is expected to provide more accurate and improved voice recognition even if users use it in a noisy environment. The upgrade also promises improved search speeds. The upgrade will also enable Google Voice Search to accurately recognize similar sounding letters and group of letters with just a few computation, according to Techno Buffalo.
Additionally, Google announced that the upgrade will also work on Android's dictation feature. Users will feel the improved performance once the upgrade is rolled out.
In order to achieve this significant increase in performance, Google introduced artificial noise and reverb to its training data which allows the system to better Google voice search users, even when they are in a noisy environment.
After the upgrade, inquiries made through Google Voice Search are now 300 milliseconds faster with a much better memory. The upgrade in memory means that the service can now hold larger snippets of audio. Prior to the upgrade, Google Voice Search can only analyze 10 milliseconds worth of audio.
Google Voice Search works in the same way as other well-known personal digital assistant just like Siri on iOS and Cortana on Windows.