EVA Voice Biometrics
Confirm people's identities and liveness with voice biometrics
EVA Voice Biometrics is a second-factor voice biometric solution designed for use with a broad range of web browsers that support audio capture. The biometric verification process is intended to be part of an authentication flow to enhance security, privacy and ease-of-use. EVA Voice Biometrics is an offering from Auraya Systems.
The voice tokens used with EVA Voice Biometrics are predominantly digit strings, although phrases are also supported. The usual verification process will usually prompt the user to read out a sequence of random digits, a sequence of consecutive digits, and optionally their phone number.
EVA Voice Biometrics is a second-factor voice biometric solution designed for use with a range of web browsers that support audio capture. The biometric verification process is intended to be part of an authentication flow to enhance security, privacy and ease-of-use. EVA Voice Biometrics is an offering from Auraya Systems.
The voice tokens used with EVA Voice Biometrics are English digit strings, although other phrases are can be supported with customization. The usual verification process will usually prompt the user to read out a sequence of random digits, a sequence of consecutive digits, and optionally their phone number.
EVA Voice Biometrics supports the most commonly used PC and mobile browsers, except where the OS restricts the audio input (e.g. Chrome on iOS). It has been tested with the following browsers:
To integrate EVA Voice Biometrics for Auth0 you will need access to an EVA Voice Biometrics server. Auraya Systems provides a evaluation service that is free to use, but is restricted to 30 day voiceprint lifetime. This free service does not make any data security or retention guarantees.
Once you have assessed the behaviour of EVA Voice Biometrics for Auth0 and you are ready to purchase your own service, please contact Auraya Systems for licensing options.
- Self managed AWS CloudFormation installation script + licence
- Auraya managed AWS service + licence
The Auth0 EVA Voice Biometrics Action must be part of your authentication pipeline. Configuration keys are used to enforce security between Auth0 and EVA-WEB server. Additionally, script keys configure which voice tokens to prompt.
Add the Auth0 Action
- Select Add Integration (at the top of this page).
- Read the necessary access requirements and click Continue.
- Configure the integration using the following fields:
AURAYA_PERSONAL_DIGITS key would require a
phone_number to be set in the
user_metadata, otherwise it would not be prompted.
|Authentication URL||EVA Voice Biometrics endpoint||
|Client ID||JWT client id on the EVA server (and this server)||
|Client Secret||JWT client secret on the EVA server (and this server). Note that this Secret will be used to sign the session token sent to EVA. The evaluation secret provided should not be used in a production environment.||
|Issuer||this application’s issuer||
|Prompt for random digits?||Whether to prompt for random digits voice token.|
|Prompt for common digits?||Whether to prompt for common digits voice token.|
|Property that contains personal digits||a user.user_metadata property that contains digits such as phone_number||
|Prompt string to use for common digits||a digit string to prompt for common digits||
|Prompt string to use for personal digits||a string to prompt for personal digits||
|Debug enabled?||if enabled, enables debug logging on console.|
- Click Create to add the integration to your Library.
- Click the Add to flow link on the pop-up that appears.
- Drag the Action into the desired location in the flow.
- Click Apply Changes.
Evaluation: Auraya Managed Evaluation Server
Use the example keys provided above (they are also the default).
The evaluation server limits the lifetime of voiceprints to 30 days.
Production Option 1: Self Managed Server with AWS CloudFormation
Contact Auraya Systems at firstname.lastname@example.org
You will be provided with access to the support portal and a licence. From the support portal you can download all the assets and AWS cloud formation installation steps to set up your own server. The installation typically takes less than 1 hour.
Production Option 2: Auraya Managed Production Server
Contact Auraya Systems at email@example.com to discuss commercial terms.
User Voice Enrolment
The first time a user is directed to EVA Voice Biometrics they will be invited to enrol their voice. This process captures between 3 and 8 utterances depending on the required voice tokens, and includes a biometric “Opt-In” consent step. At the conclusion of enrolment they will be redirected back to the authentication flow.
At the start of the process , the browser will prompt the user to allow access to their microphone - for most browsers this is a one-off step. The user is also able to select a microphone from a dropdown, and can trigger audio capture either by clicking the microphone icon (or anywhere within the widget border) or by holding down the space bar.
A successful enrolment will set the following user metadata:
user.user_metadata.auraya_eva.status = 'enrolled';
A user’s enrolment can be reset by deleting the
user.user_metadata.auraya_eva.status property or by setting to the value
User Voice Verification
An enrolled user is redirected to EVA Voice Biometrics where they are prompted to say between 1 and 3 utterances depending on the required voice tokens. The user is given two tries to complete each token. Once completed (either successfully or not) the user is redirected back to the authentication flow.
The configured voice tokens at verification time must have previously been enrolled, otherwise the user will be rejected.
Verification success rates can be affected by a change of microphone or background noise. The process works best when
- the enrolment and verification environments are quiet
- with close lipped microphones or noise-cancelling array microphones
- using the same microphone for enrolment and verification