VoiceBase: Building a global speech analytics platform with Google Cloud Speech API

San Francisco-based VoiceBase provides a speech analytics platform that turns voice data into useful information that can help companies automate and gain insight into their operations. The company transcribes and processes millions of recordings every month for contact centers, conferencing service providers, video and education platforms, telecom providers and Fortune 500 companies. The service goes beyond providing transcripts and offers tools to solve complex business problems — such as analyzing audio files and transcripts from call centers to predict the level of customer churn, and letting users search the timeline of a recording to find and play precise parts of a conference call, webinar, educational lecture, podcast or video interview. VoiceBase wanted to extend its services globally, so files could be automatically transcribed regardless of the language spoken.

“I started VoiceBase because there was no way for businesses to easily search, share and store audio information. In the same way that email and rolodexes have moved to the cloud, we decided to move voice data to the cloud and help companies mine it for usable information. To go global with our service we chose Google Cloud Speech API to automatically transcribe audio recordings in 80 languages and dialects.”

— Walter Bachtiger, Founder and Chief Executive Officer, VoiceBase

Turning audio into usable information

Audio recordings, such as those made in call centers, frequently are available as archives but otherwise stored and forgotten. That means that valuable information that could be useful to the company, customers and others is completely overlooked because it’s in audio form that is not easily searchable or otherwise analyzed without replaying it. With VoiceBase, spoken information can be mined for business insights. The software has built a speech analytics platform that tags and analyzes meaningful keywords and phrases that can be used to identify sales leads versus non-prospects, detect potential churn, and monitor the behavior of call center agents.

To do that, VoiceBase automatically transcribes voice files, performs analytics on the transcripts, and uses deep-learning techniques to extract insights from them. The quality of the insights is dependent on the quality of the transcript — without accurate transcriptions, little usable information will be produced. VoiceBase chose Google Cloud Speech API to transcribe recordings in multiple languages and offer the broadest language coverage possible for its customers, including dozens of languages its platform had not previously supported.

“We work with large, global brands, and one of our challenges as a smaller company has been providing them with the global language coverage they need. Google Cloud Speech API provides world-class language coverage at an affordable price. The accuracy of Google Cloud Speech API transcription ensures that we provide our customers with the highest quality of service possible,” Bachtiger says.


Expanding the business by covering more languages

VoiceBase’s use of Google Cloud Speech API is still in the beta phase, and has not yet been introduced commercially. When it goes live with coverage of 80 languages and dialects, VoiceBase will be able to target global customers that it previously could not serve because of the companies’ multiple language requirements. VoiceBase is a small company and could not on its own afford to create tools to automatically transcribe speech from multiple languages.

“As we engage with larger, multi-national companies, there’s a growing need for our platform to work with more and more languages. Google Cloud Speech API provides that for us at an affordable price. Its broad coverage allows us to quickly scale globally. With it, our customers can be confident that our transcriptions are accurate,” Bachtiger says.