COVID era is all ears to voice biometrics

The 1968 Hollywood film In 2001: A Space Odyssey portrayed a computer named HAL 9000 that was shrewd enough to authenticate by voice, comprehend speech, and discern emotions. Likewise, Star Trek made in 1966 showcased the proficiency of the spaceship’s computer to identify the crew member and also distinguish between voice commands and conversations between the crew members. Biometrics used in recent sci-fi movies like Dredd, District 9, Ex Machina etc. are depicting a futuristic world of groundbreaking technology.

The current coronavirus pandemic, which is suddenly sweeping the planet, has made its inhabitants so spooked of touch that it has swivelled the spotlight on voice biometrics. As both the routinely existing techniques of biometrics, touch (fingerprint), and the facial recognition technique carry the risk of amplifying the dissemination of the coronavirus. Touch is one of the most fundamental ways by which coronavirus gets transmitted hence relying on biometrics which hinges on contact could spell doom. Second, though facial recognition systems are contactless, it drastically curtails the precision of the system when a person is wearing a mask. Further, urging a person sporting a mask to take it off would not only jeopardise him but others as well, by fostering the proliferation of the virus, especially when talk of the second wave is doing the rounds.

Voice has therefore emerged as a feasible authentication method because, just like a fingerprint or a face, a voice is unique to an individual, and a biometric voice engine facilitates validation of a person’s identity by his or her distinct vocal-characteristics. Voice biometric systems work by generating a voiceprint or ‘template’ of a person’s speech by capturing the many peculiarities of a person’s voice, including the sound pattern, tempo, cadence, inflexion and rhythm in the form of a ‘Voiceprint’. A voiceprint so created gets encrypted and stored for future voice verification. A voice-based biometric system matches the voice of the person to the ‘Voiceprint’ in the database, authenticating him or permitting or denying him access. Voice biometrics facilitates quick, frictionless and highly secure access and enhances verification experience by eliminating frustration associated with lost and stolen credentials and clunky login processes. Because of lower execution cost, ease of use, and greater accuracy, voice biometric technology is emerging as a game-changer.

When compared to other forms of biometric applications, voice biometrics comes off as the choicest one for non-intrusive, contactless and easy-to-use systems. A study carried out by Unisys disclosed that the consumer propensity for voice recognition was highest at 32 per cent, followed by fingerprint 27 per cent, facial scan 20 per cent, hand geometry 20 per cent, and iris scan 10 per cent. This ranking confirms that convenience and experience appear to be the determining facets when choosing a biometric technology. According to Opus Research, there will be over a half-billion voiceprints by 2020. Google, in 2017, divulged that their speech recognition had an accuracy rate of 95 per cent. Further, Google has now disclosed that 20 per cent of their searches are being made by voice query, which they expect to grow by 50 per cent by 2020.

The voice recognition technology has both advantages and disadvantages. Voice commands are a very efficient tool; it can boost productivity levels as talking can get done faster than typing. Voice recognition software is disadvantageous as it may not put words accurately on the screen; besides, there could be interference because of background noise and non-recognition of some accents. Voice hacking in voice biometrics is not far-fetched, an attacker can spoof an individual’s unique voiceprint to steal his identity, and access voice-based systems. Advances in machine learning, recording technology and synthetic speech are enabling high-quality voice spoofing, showing that hackers can trick voice-based systems. Therefore, it is best to use voice in combination with other forms of authentication such as behavioural biometrics, geo-location, passcode, fingerprint or an ID. AI and natural language processing if harnessed and blended could make voice authentication even more secure and reliably a fingerprint of the future.

Voice biometrics technology is also an Audio Forensic tool to analyse voice in crime prevention and investigation efforts. Forensic experts are today in a position to detect samples of speech in an audio recording easily and analyse hundreds of audio files in a few minutes to identify individuals in an instant no matter what their accent, language or content is—in the process, it’s saving several thousands of hours in indicting suspects. We have included the voice-biometric device in the list of equipment to be installed at the new cyber lab proposed to be set up at a cost of Rs 6.3 crore at our directorate for the examination of audio files. The judiciary, too, has been favouring the voice-based systems in its various orders. The Supreme Court on August 2 last year held that a judicial magistrate could direct a suspect to provide his voice samples for investigation even without his consent. A three-judge Bench, led by Chief Justice of India Ranjan Gogoi, ruled that directing a person to part with his voice sample to police was not a violation of his fundamental right to privacy. Similarly, in the Mahabir Prasad vs Surinder Kaur case, the Supreme Court ruled that police could use the tape-recorded conversations as corroborative evidence.

In a post-pandemic world, moving from a touch interface to voice wherever it’s workable will be a necessity. Voice is already in use as a biometric for smart speakers such as Google Home, where the Google assistant recognises the voice of the users who seek information from it.

Lloyds Bank, TalkTalk, Vodafone, have already rolled out voice-based services, which is letting people access their accounts without pin or passcodes by using their voices. Others like the BBC, Flipkart, Snapchat etc. have launched voice assistants for the convenience of their users. Organisations can use voice biometrics to ensure that they not only render excellent service to genuine callers but also to prevent suspected fraudsters from perpetrating fraudulent activities in real-time. If a person loses his credit card, calling the bank to get the card blocked is tedious. But, if the bank has a biometric voice system, it can complete the verification the moment the customer calls by comparing the caller’s voice with the voiceprint available with the bank in no time.

Finally, if we accept that we are spiritual beings having a human existence, wouldn’t our ‘spiritual identity’ or ‘soul-based biometrics’ which leans on our consciousness instead of voice be the best system to outwit the coronavirus of the future?