Voice-based Age Estimator
Date: November 2023
Duration: 1 week
Team: 2 Product Designers - worked on tasks together
Tools: Figma, Miro, Mobbin
.png)
Company: ParallelChain
ParallelChain Lab is a blockchain and AI software company, leading the digital identity revolution with solutions for web2 and web3.
Project Overview
The task for the design team was to create an interface for a product which uses AI technology to detect the age of the user from their voice recording.
We had won the IMDA/PDPC Challenge Call for Age Estimation, jointly organised by the Singapore Infocomm Media Development Authority (IMDA) and the Personal Data Protection Commission (PDPC), the national authority for data protection matters.
Problem Statement
With the rise in data protection from governments, there is also a need to find alternative solutions to verifiying identity that does not involve capturing faces, especially children's. The solution is to create a product that can detect the age of the user using their voice, and does not require capturing sensitive data. The company had entered this competition and has been working on this product as part of the digital identity verification suite.
Design Brief:
-
Step one - agree with terms and conditions
-
The user would be reading a sentence into the device
-
A mic button to be used for the recording
-
Timer for security purposes for user to complete the task in the time given
-
The webapp should be supported on mobile/desktop
Use case examples:
gambling website, buying items or accessing websites that require age verification
Design Process
I used a Design Thinking approach for solving this problem. Following the order of user research, ideation, prototyping, testing, and iteration. However due to the time and budget constraints, we were limited to surveys and 2 longer interviews during the research process.
Research
For the research phase, due to time and budget constraints we first issued short surveys then made 2 longer interviews with participants. We analysed how users would use apps that used a voice recording element, as there are not many similar applications. In addition we looked at existing softwares or pages to see the components used in verifying users.
Survey Key Findings
Survey results of 6 participants between the ages 22-38 years old.
-
Current satisfaction score avg. 7.5/10
-
Frustration score avg. 3.8/10
-
Likelihood to recommend improvements avg. 8.5/10
-
83% of participants trust voice AI technology
Although th current satisfaction scores are good, and there is a low frustration score, there is still scope to improve as the likelihood to recommend improvements averaged 8.5/10. It would be useful to look at the existing user flows.
Longer interview key points:
-
Privacy and data: participants did not want to submit passport of photo ID
-
Visual: wanted visually appealing clean interface
-
Accessibility: text can be hard to read in different fonts
-
Technical issues: loading errors and having to refresh the page and losing the information of a completed form
-
Time consuming: takes long to complete or having to repeat stages
Thoughts about how to incorporate findings into design
-
From analysing the data above, the AI voice age estimator is a good option for users not wanting to submit sensitive information such as their identity.
-
We can also use a sans-serif font to help with accessibility and voice capture would be potentially faster for some users. This would also be useful for those who do not have their identification at hand.
-
For addressing the technical issues, we could implement a button that allows the user to refresh for another phrase rather than reloading the page.
Competitive Analysis
This stage included exploring the market, looking at direct and indirect competitors on age verification.
Findings:
Many current age verification technology requires users to enter all their details of their ID etc and which could be time consuming. There are also options for scanning ID which is a faster process.
Amazon Prime
Pros:
-
step-by-step instructions
Cons
-
long, time consuming to complete
-
errors in recognising age in databases
-
user may not have their ID card at hand
The interface is much simplier and easy to use in this example from Roblox.
The CTA button is very clear and has simple process.
Roblox
Google Translate
.png)
I analysed the design in google translate is one of the top downloaded apps using voice technology.
User Personas
The user base would cover a wide range of users, as this software is used to identify and verify the age of users. In this instance we would focus on a scenario, e.g. a user trying to access a gambling website.
David
David is a 28-year-old marketing professional who watches action films. He needs to verify his age to watch age restricted films.
Goals and Needs:
-
wants an easy way to verify age
-
does not want to fill in credit card or ID details - thinks it's time consuming and worried about giving sensitive data
-
does not always carry around his ID
Age: 28
Gender: Male
Occupation: Marketing Professional
Location: New York City
Task Flow
Next, I created the main task flow to understand the flow of the user and to find any edge cases. This shows how the user would start and record their voice then the final result.
.jpg)
Design Phase
Colour
Choosing from the selection of the brand colours in the design system, we chose the "brand orange" for our primary colour to follow our branding, and to differentiate from other existing products in our suite. We then used a mixture of greyscale colours to create the colour palette.
A/B Testing
Iterations
-
changed the terms and conditions box smaller as the tick to agree and continue was not visible in the window and user did not know how to continue
-
made the instructions text larger and showed all steps on the page, as user did not see and did not know they had to read the text on screen
Updated Task Flow
-
removes the step to click next after the instructions
-
adds in the questionnaire required after reading the text
Reflections
-
Design sprint lasting 1 week
-
Identified challenges: creating a clear interface for voice-based age detector
-
Conducted user research to understand age verification and pain points
-
Competitor analysis to understand the existing features - incorporated microphone and stop sign in final design
-
Developed user personas and journey maps, highlighting pain points and goals.
-
Prototyped design concepts, iterating based on usability testing feedback - we did A/B testing as well as user testing - preventing errors
-
Balancing user insights with technical constraints and project timelines, as well as business needs or requirements from the management team
Contributions
-
Mainly worked on contributing to brainstorming ideas, leading the survey research and analysis, adding design system, A/B testing and iterations
What I would do differently
-
If we had more time and a budget I would have expanded on the research and done more longer interviews to get more insights on users
-
done more user testing - I think to improve the product it may be better to include pop up prompts directing the user where to go on the page
Next Steps
-
The next steps for this project should follow more user testing and iterations to the product.
-
we can use software such as lookback to observe users and improve user flows
-
we can also utilise heatmaps to see how users scan the page - and iterate designs accordingly
-