Dark Reading is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Endpoint

2/20/2017
11:45 AM
Connect Directly
Twitter
RSS
E-Mail
50%
50%

Voice Biometrics Prone To Error, Study Shows

New research shows the need for a holistic solution to authentication, not just voice biometrics alone.

It's hard for the untrained ear to notice, but the human voice changes over a period of several months and years, primarily the voice's pitch and speed of speech.

Researchers at Pindrop have discovered that error rates on voice biometrics may double in a two-year span, which is why they say companies need to consider multifactor authentication when setting up voice authentication systems.

Representatives from Pindrop presented their findings last Friday at the annual RSA Conference in San Francisco.

"Too many companies will just deploy voice biometrics, and that’s a mistake," says Terry Nelms, director of research. "There’s a real need for multifactor authentication, with voice biometrics just being a piece of the puzzle."

Elie Khoury, principal research scientist at Pindrop, says the team studied former-President Barack Obama's speech patterns in his daily addresses from January 2009 to January 2017 and found that the former President's voice-accuracy rating dropped by 23%.

The Pindrop research team also tracked 122 speakers in six languages (English, French, German, Spanish, Dutch, and Italian) and error rates doubled from 4%- to 8% in two years. One interesting note: The female voice does not change as much as the male voice, and voices of men aged 60 and over change the most dramatically.

"When people speak, they use up to 100 muscles," explains Khoury. "So like any muscle, they become less flexible with age.  Similarly, the vocal cords and cartilages of the larynx also change as we age."

Khoury adds that the research found that 48% of callers only call in to the average banking call-center system once every eight month, and the human voice can change over that time.

To counteract changes in the human voice, Pindrop uses compensation techniques that take into account the caller's gender, age, and the time interval between the last call and the next call.  In addition, it's possible to update the model over time, but before any update Pindrop recommends applying a multifactor authentication technique such as phoneprinting to prevent identity theft. 

Frank Dickson, an analyst at IDC who covers security and authentication, adds that authentication is not a binary event.

"Authentication is not a switch, it’s a score," he explains. "What voice biometrics does is improve the probability that the company is talking to the right person."

Most financial institutions use the passive approach, in which the caller doesn't have to be actively enrolled. Their speech is processed and verified during their normal interaction with the call agent. Pindrop's Khoury says the level of accuracy depends on the amount of net speech; however, a few seconds of normal conversation is typically enough to authenticate a user or detect fraud.

Pindrop uses the combination of voiceprints and phoneprints to accomplish multifactor authentication.  Phoneprinting analyzes a call to extract features from the audio revealing call type, geolocation, and the unique phone to authenticate the caller and detect fraud. 

Related Content:

 

Steve Zurier has more than 30 years of journalism and publishing experience, most of the last 24 of which were spent covering networking and security technology. Steve is based in Columbia, Md. View Full Bio
 

Recommended Reading:

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
COVID-19: Latest Security News & Commentary
Dark Reading Staff 9/25/2020
Hacking Yourself: Marie Moe and Pacemaker Security
Gary McGraw Ph.D., Co-founder Berryville Institute of Machine Learning,  9/21/2020
Startup Aims to Map and Track All the IT and Security Things
Kelly Jackson Higgins, Executive Editor at Dark Reading,  9/22/2020
Register for Dark Reading Newsletters
White Papers
Video
Cartoon
Current Issue
Special Report: Computing's New Normal
This special report examines how IT security organizations have adapted to the "new normal" of computing and what the long-term effects will be. Read it and get a unique set of perspectives on issues ranging from new threats & vulnerabilities as a result of remote working to how enterprise security strategy will be affected long term.
Flash Poll
How IT Security Organizations are Attacking the Cybersecurity Problem
How IT Security Organizations are Attacking the Cybersecurity Problem
The COVID-19 pandemic turned the world -- and enterprise computing -- on end. Here's a look at how cybersecurity teams are retrenching their defense strategies, rebuilding their teams, and selecting new technologies to stop the oncoming rise of online attacks.
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2020-15208
PUBLISHED: 2020-09-25
In tensorflow-lite before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1 and 2.3.1, when determining the common dimension size of two tensors, TFLite uses a `DCHECK` which is no-op outside of debug compilation modes. Since the function always returns the dimension of the first tensor, malicious attackers can ...
CVE-2020-15209
PUBLISHED: 2020-09-25
In tensorflow-lite before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1 and 2.3.1, a crafted TFLite model can force a node to have as input a tensor backed by a `nullptr` buffer. This can be achieved by changing a buffer index in the flatbuffer serialization to convert a read-only tensor to a read-write one....
CVE-2020-15210
PUBLISHED: 2020-09-25
In tensorflow-lite before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1 and 2.3.1, if a TFLite saved model uses the same tensor as both input and output of an operator, then, depending on the operator, we can observe a segmentation fault or just memory corruption. We have patched the issue in d58c96946b and ...
CVE-2020-15211
PUBLISHED: 2020-09-25
In TensorFlow Lite before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1 and 2.3.1, saved models in the flatbuffer format use a double indexing scheme: a model has a set of subgraphs, each subgraph has a set of operators and each operator has a set of input/output tensors. The flatbuffer format uses indices f...
CVE-2020-15212
PUBLISHED: 2020-09-25
In TensorFlow Lite before versions 2.2.1 and 2.3.1, models using segment sum can trigger writes outside of bounds of heap allocated buffers by inserting negative elements in the segment ids tensor. Users having access to `segment_ids_data` can alter `output_index` and then write to outside of `outpu...