Google announced that Google voice search app on Android and iOS is now improved in voice input. The app is now intensive, faster, and gives accurate results even at noisy environments.

The company is perhaps the most low key about using voice controls and interaction in their software. Its Now voice assistant does not personify unlike Microsoft's Cortana and Apple's Siri. But among them, it is the best voice recognition algorithms because it can easily recognize mumbled inputs with almost no processing delay. The company is even claiming that it can be robust to noisy environment.

Like Us on Facebook

Before, the system could analyse 10-millisecond snippets of audio. It makes predictions of words on the basis on the recognized sounds regardless of the order they are uttered. Now, the system has a better memory. It can consume larger snippets of audio. It also gets concern to itself on the order of sounds that were spoken. Using Connectionist Temporal Classification and sequence discriminative training techniques, Google voice search tool will now be built on better neural acoustic models according to The Next Web.

The voice search engine can predict phonemes quickly and differentiate between similar sounding letters using this new technology. It is now better at analysing every snippet of audio by referring to the sounds on either side of it, making it faster, more accurate transcriptions, and efficient voice search experience in noisy places. Its aural wizardry will also make the dictation on Android devices better with less computing power. Deep neural networks that make freaky distorted pictures have been employed to compute the words said to a phone, but now it is evolved in using recurrent neural networks. Recurrent Neural Network (RNN) is the improved system and allows temporal dependencies.

In a new post of Fortune, Google Speech Team members have added the latest developments in the company's voice search algorithms and added ambient noise and reverb to the data it used to train its new system.

TagsGoogle Voice, Search engine