2 min

Google has announced its Artificial Intelligence (AI) technology behind its Contact Center AI service to update to improve voice recognition. Contact Center AI is based on Dialogflow and Cloud Speech-to-Text.

The first update is for Dialogflow, writes Silicon Angle. Dialogflow is a cloud-based service created on the basis of the acquisition of API.AI in 2017. Dialogflow allows enterprises to create virtual assistants without having to build complex machine learning technology from scratch.

The tool now gets a whole new feature called Auto Speech Adaptation, which is now launched as a beta. The function is designed to improve the contextual awareness of virtual agents. In this way, chatbots should better understand speech and respond appropriately by taking the context of the conversation into account in their choices.

If the Dialogflow agent knows that the context is ordering a citizen and that cheese is a common ingredient of citizens, then it probably understands that the user intended cheese (cheese) and not this (thesis), explain product managers Dan Aharon and Shantanu Misra.

Cloud Speech-to-Text

Cloud Speech-to-Text will also be updated. This is Google’s transcription service, which uses machine learning to accurately convert human speech into text form.

For developers, the service includes a tool called SpeechContext parameters, which can be used to add contextual information to improve the accuracy of transcriptions.

SpeechContext will now be updated to allow it to add classes – or popular concepts – to virtual agents. In this way they should get a better understanding of sentences that can be confusing.

New possibilities

SpeechContext also has new features, including a boost function that helps to reduce the number of false positives. A false positive occurs when a sentence was not pronounced, but does appear in the transcript.

Finally, the phrase hints feature helps developers to reduce the chance that common words and phrases in their company are well understood and described by Cloud Speech-to-Text.

This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.