Voice Security Company Pindrop Raises $90 Million, Sets Sights on Securing Smart Speakers and European Expansion
Pindrop, the call center security company that uses voice analysis to detect fraud announced last week that it had raised $90 million in new funding from Vitruvian Partners, GV (formerly Google Ventures), Andreessen Horowitz, Goldman Sachs, and six other firms. Pitchbook reports the post-funding valuation is $900 million. Pindrop plans to use the funds to expand into Europe and new security product categories outside of call centers such as voice-activated IoT devices, including smart speakers. According to the announcement:
“The new infusion of capital will also serve to advance Pindrop’s evolution towards providing security and identity solutions for voice-assisted smart devices ranging from Google Home to smart locks to connected cars. Pindrop’s deep roots in securing the call center enables the company to apply its advanced deep learning technology to make voice a trusted form of biometric authentication for voice-enabled devices.”
The Rise of Voice Security and Biometrics
Pindrop’s business is focused on fraud detection in call centers and a key component is voice biometrics. The company intends to extend those capabilities to IoT devices, including smart speakers, as the device makers and their voice assistant platforms allow. Earlier this year, Pindrop posted a video that simulates an Alexa skill for Bitcoin transactions that is secured by voice biometric technology. This is similar to the technology used by Google Assistant and Alexa to identify who is speaking. However, the example is applied to a simulated third-party Alexa skill. It shows how the account owner can be recognized by voice and transactions are allowed while another speaker cannot.
Voice Biometrics Face Obstacles on Smart Speakers
Voice biometrics are used today in a variety of consumer use cases including mobile app and smartphone login. Sensory announced a new solution for smartphones and computers earlier in 2018 called TrulySecure Speaker Verification 2.0. This performs a similar function of voice verification associated with a user login to devices or apps. As of May, it was already in use on devices from Samsung, LG, and Motorola. Smartphones are touch-first and voice-second devices and voice biometrics are still considered a valuable feature. As voice-first devices, smart speakers can extract even greater utility from voice biometrics since authentication often can only be validated by using spoken inputs from the user.
The rapid growth of smart speaker adoption and rising sales of other voice-activated devices surely makes this an attractive market for Pindrop and other technology providers with voice biometric capabilities. Both Norton and McAfee are predicting that 2019 will bring a rise in cyber attacks targeting consumer IoT-device and smart speakers in particular.
However, the type of voice biometrics offered by Pindrop and Sensory currently cannot be implemented through Amazon Alexa and Google Assistant. Neither platform provides audio files of user utterances to third-party solutions. Only a speech-to-text transcription of the utterance is provided, or sometimes merely the intent interpreted by the voice assistant. As a result, any service that requires the audio of the speaker cannot be implemented today. That may well change in 2019, but don’t expect a lot of Alexa skills or Google Actions with voice biometrics features anytime soon, unless the validation is actually performed by Alexa or Google Assistant directly.