PArse - UX Researcher
How Can We Make Voice Assistants More Accessible?
Background
Speech Disorders
5% to 10% of Americans may have communication disorders
By the first grade, roughly 5% of children have noticeable speech disorders.
3 million+ Americans stutter.
Nearly 7% of Americans have some form of language impairment.
Approximately 1 million Americans suffer from aphasia.
Lack of Accessibility
No major voice assistants have comprehensive accessibility options for persons with moderate/severe speech impediments.
Voice assistant technology has existed in some consumer form for a decade without a viable accessibility option.
Users with speech impediments are not being properly considered in the design process
These users are victims of the technique of “designing for the average”, not taking into account users with differing, unique speech patterns during product design.
These users should not be neglected and in fact, should be directly involved in the product design process.
Research Phase 1 - Generative
Research Questions
How do persons with speech impediments currently use voice assistant technology?
What problems do persons with speech impediments face when using voice assistant technology?
What type of solution would these users desire to solve these problems?
Research Goals
Identify the scenarios, environments, and general context in which users with speech impediments use or would like to use voice assistant technology.
Identify the major problems that persons with speech impediments face when using voice assistant technology.
Identify possible solutions to these problems to explore during participatory co-design.
Research Methods
Competitive Analysis - Answer questions such as: What voice assistant technology is currently on the market? What accessibility options currently exist for persons with a speech impediment? Who are our primary, secondary, and tertiary competition? What hardware currently exists to act in tandem with voice assistant technology? How much does this physical hardware cost? What needs are voice assistants currently not filling? What do users think of these voice assistants in reviews and feedback? How are these companies engaging with users over social media? Are there any particular features that make their product unique relative to other similar apps?
Interviews - In-depth interviews would fill in our massive knowledge gaps from actual users with speech impediments to understand not only their usage of voice assistants but try to understand their unique life experiences better. These interviews would be moderated, semi-structured, and last around 30 minutes, the other 30 minutes would be dedicated to the contextual inquiry. The open-ended questions would concern participants’ life experiences, their experiences with their speech impediment, their usage of voice assistants, what they seek from voice assistants that they aren’t currently getting, and how these experiences could be improved.
Contextual Inquiry - In addition to user interviews, we also wanted to see how users interacted with voice assistants. During the contextual inquiries, we asked users to use a voice assistant they had previous experience with and asked them to do whatever actions they thought were enjoyable or useful. We recorded these with video and took notes, while also asking probing questions.
Sample Size - three participants
Demographics - two self-identified men and one self-identified woman, between the ages of 25 and 39. Two participants live in Washington and one participant lives in Texas. One is a student, all three are working professionals.
Inclusion Criteria - users with speech disorders/impediments
Exclusion Criteria - users without any experience or knowledge with using voice assistants
Analyses/Synthesis
Competitive Market Analysis - The competitive market analysis will represent the most popular rival apps, their features, their critic rankings, their audience rankings, their total profits, and other information.
Affinity Mapping - the data from this study will be primarily qualitative. The researcher will have to organize the collected data to obtain the design team's critical insights. For tasks like the participatory group design, key features will be represented by cards and ranked based on broader categories. The researcher logs critical elements across all the tasks, and the vital features with the most significant consensus will be highlighted and presented. Graphs can help represent these key features, especially ones that are common to most participants.
Main Findings
Voice Assistants are currently unusable most of the time for these users - all three participants don’t currently use voice assistants due to the technology not working on a regular basis or at all in certain cases.
Persons with speech impediments feel “left behind” by technology - all three participants felt like the technology “wasn’t meant for them” and that they are being “left behind” by major tech companies.
Persons with speech impediments have similar experiences when socializing, speaking - each participant had unique life experiences, but all three described common occurrences of being interrupted, ignored, spoken over, and generally neglected in conversation with others.
Each person with a speech impediment has a unique speech pattern - the participants noted that each person with a speech impediment has a different speech pattern, with different quirks, tics, pauses, extensions, etc.
Users want to feel like they are part of the technology boom - users want to feel connected with society at large who are currently excited about technological developments regarding voice assistant technology.
Users don’t want an entirely new voice assistant - users were less interested in a completely new voice assistant solution, and more in a method for making currently available voice
Research Phase 2a - Ideation
Research Questions
What features do persons with speech impediments value when using current voice assistant technology?
In what format would persons with speech impediments want these features?
What additional features would persons with speech impediments like in a voice assistant technology that isn't currently offered?
Research Goals
Identify which features persons with speech impediments value when using current voice assistant technology.
Identify an acceptable format to deliver these features to users.
Identify which additional features persons with speech impediments desire in a voice assistant technology that isn’t currently offered.
Research Methods
Participatory Co-Design
Main Findings
Users want a seamless experience between multiple types of hardware - Voice assistants have applications beyond the phone. Users want to be able to use their voice assistant for tech home applications, Bluetooth speakers, etc.
Users want to be able to use an existing voice assistant technology - Reiterating what users mentioned in our original interviews, users want to be able to use the same voice assistants that everyone else uses.
Users want the device to be able to “know” them - Users want to feel like the system was designed for them, not that they are getting the leftovers of a design for everyone else.
Users want to be able to customize settings based on their unique speech patterns - Each user has different speech patterns, with different difficulties, the user thought it would be great to control some specific options to make their experience easier depending on these qualities.
Users want to feel like their privacy is safe = Users feel vulnerable when using an accessibility app since they don’t like to draw attention to their disorder so they would like to make sure that their personal information is protected.
Users have no major preference in the format as long as it satisfies their conditions - users had no real preference between phone app, wearable, earbud, wireless speaker, as long as it fulfilled their goals of being able to use an existing voice assistant.
Research Phase 2b - Iteration
Research Questions
Do users with speech impediments find our design prototype useful?
Do users with speech impediments find our design prototype desirable?
What improvements can we make to our design prototype?
Research Goals
Identify modifications that we can make to improve our design prototype based on user feedback and direction
Research Methods
Iterative Feedback/Interview
Main Findings
Our solution was exciting for participants - Participants were really excited at the prospect of our solution being implemented.
The features were both useful and desirable - The features and functions made sense to them and were features that they would use on a consistent basis.
The interface designs were well-received - The overall design of the prototype was liked by participants.
Some language could be tweaked to be more inclusive - one possible issue is that some of the wording that we used wasn’t as inclusive as participants would like. For example, our prediction accuracy in percentage was labeled as “confidence level”, which participants suggested should be changed to something like “prediction accuracy”.
More privacy information would be desired - Since the prototype is an app that includes transcription with an audio device, users wanted more information about where and how their information was stored, as well as when it would be deleted.
Research Phase 3 - Usability
Research Questions
How usable is our prototype?
Are the language and content inclusive for persons with speech impediments?
Are there any further design modifications that we should make to our prototype?
Research Goals
Identify modifications that we can make to improve our design prototype based on user feedback and direction
Research Methods
Qualitative Usability Testing
Main Findings
Our solution was overall usable
We fixed inclusive language issues
Our privacy information was comprehensive