VoiceAssure: A ROBUST, PRIVACY-FIRST, VOICE-BASED AGE ESTIMATION TECHNOLOGY
Main outcomes
- Privately received an independent , UK regulator approved prestigious Level 1 certification and joins a handful of global companies to deploy their age assurance products across many industries.
- VoiceAssure is a voice-based privacy-preserving age estimation technology with liveness and quality checks in place.
- VoiceAssure is efficient in keeping minors <15 out of adult spaces while providing low-friction access to adults.
- Voice-based age estimation brings low-friction performant age checks for many use cases like advertising, website visits and audience measurements.
- VoiceAssure has already been deployed in restricted content industries since June 2022.
- Privately’s ML infrastructure allows them to continue building out their products at high speed and reliability.
About Us
Privately is a Swiss company deploying online safety technology since 2014. We help platforms deliver age appropriate and safe online experiences to children and adults alike. We democratize age estimation for developers, and automate regulatory compliance for companies.
Our on-device AI technologies determine age, well-being and online safety markers of the child through text, image, voice and behavior analysis. Our technologies are privacy-preserving, and allow real-time support and interventions to be integrated within the user experience without the need for any data to leave the device.
The technology can be integrated within our clients’ apps, games or devices running privately within a smart-device environment through our Software Development Kit (SDK) or via secure on-browser implementation.
What is Age Estimation?
Age Estimation technologies deduce the age range of a person based on the analyses of biometric features, such as facial patterns or voice patterns. Though relatively new, the automated detection of age promises to revolutionize many industries and make the internet a safer place for minors.
Historically the purchase of restricted goods and services has been subject to age verification in the following retail spaces:
- Alcohol, cigarettes, including E-Cigarettes and vaping products
- Lottery tickets and scratch cards
- Dangerous Weapons, e.g., crossbows and knives.
We now observe a regulatory push for age assurance in online worlds as well. This is on account of rapid spreads of online advertisement, social media, adult sites, online video games, augmented & virtual realities. As a response, new regulation in the UK and US mandates Age Appropriate redesign for all web services including retail. Sellers and advertisers must ensure the user’s age beyond self-declaration.
Traditionally, age estimation is undertaken by natural persons, in a wide range of settings, all of the time. These natural persons make a judgment, based on a person’s appearance, whether they are old enough to purchase age-restricted goods, content or services. However, this approach notably fails to respond to important challenges:
- Humans estimations of age can vary very significantly from real ages
- How to prevent underage access to restricted goods / content online?
- How to verify many adults who do not have or don’t want to share identity documents online?
- How to handle data processing liabilities GDPR / biometric data for underage users?
As such, AI-Assisted, On-Device age estimation technologies prove to be a vital solution for ensuring age appropriate experience.
Last year, we built and certified FaceAssure, which analyzes facial patterns to give an age estimation without ever sending any personal data to a remote server. We have noted that in some of our use cases, a voice-based age estimation would be preferred more than a facial age check. As a consequence, we have also built VoiceAssure, which is perceived as a more light-touch way to ensure that goods and services are rendered age-appropriately.
VoiceAssure analyzes patterns of speech in order to estimate the ages. It does so with complex statistical rules derived via Deep Learning methods. On top of the core age estimation routine, VoiceAssure contains other useful modules for liveness & quality check for optimal outcomes.
Making VoiceAssure Robust
Voice-based age signals are tricky to capture. In particular, we identified the following factors in audio clips that pose challenge to the robustness of age estimation during operational use on the edge, namely:
- Languages, accents and gender
- Tester’s distance from microphone, speed of speech, …
- Microphone quality and background noise
Towards this end, we have performed a continuous and rigorous vulnerability analysis and improved the performance of the system with:
- Identifying problematic inputs and edge cases.
- Identifying efficient data collection strategies.
- Identifying the best training- and test-time augmentation strategies to increase robustness to real-world perturbations.
Genuineness check
Just like in Face-based checks, we surveyed possible vectors of presentation attacks. We identified the following were the most frequent types of presentation attacks:
- Robotic voices (Siri, Alexa, Google Home, etc. being the most frequent ones)
- Pre-recorded audio clips of individuals other than the tester
- Voice acting and other voice modulation attempts
We have developed the appropriate counter-measures in the form of active and passive liveness checks. Today we’re able to cover more than 99% of the spoof attempts.
Performance numbers
With iterative developments on age estimation models, we were able to bolster our system performance to the following levels:
Independent Testing: In March 2023, ACCS, a UKAS accredited conformity assessment body, examined VoiceAssure and noted the following:
- A 100% accuracy in predicting 13–14 year olds as minors.
- A 100% accuracy in predicting 26–27 year olds as adults.
Internal Testing: We have conducted live tests with more than 600 individuals whose native language was not English but French, German, Turkish, Hindi, and so forth:
- A 100% accuracy in predicting less-than-15 year olds as minors
- A 100% accuracy in predicting older-than-24 year olds as adults
- Age estimation results were consistent when the testers spoke their native language and when they spoke English.
The system is also robust against biases on account of different genders and accents therefore very well adapted to large-scale market deployment.
Future Outlook
Our technology has a wide variety of applications, from age gating to compliance automation and age-appropriate advertisement. VoiceAssure’s successful certification further boosts our offerings to our clients, some of which were already enjoying FaceAssure’s capabilities.
Our next step is delivering a full PAS-1296 age verification solution for our clients. Stay tuned!