Mimik Studio: Automating Phonetic Transcription for Speech-Language Pathologists
enFocus | Innovation Intern
May 2024 - August 2024 • 12 week timeline • UX Designer
Problem
Speech-Language Pathologists spend significant time manually transcribing patient speech into the International Phonetic Alphabet, leading to fatigue, errors, and reduced focus on client treatment. Mimik aims to automate this workflow, but needed clear, validated user flows to move toward development and clinical trials.
Goal
Design a complete, high-fidelity prototype of Mimik’s end-to-end user experience preparing the team for usability testing and clinical trials.
Tools:
Figma
Jira
Methods:
High-fidelity Prototyping
Heuristic Evaluation
Information Architecture
Usability Study Design
Product Management
Project Impact
Designed the high-fidelity prototype for Mimik by collaborating with the founder and engineers, and consulting previous research documentation preparing it for clinical trials and user testing
Closed gaps in the end-to-end user experience by designing features that would enhance user control, allowing them to make edits and corrections that would improve the LLM and build trust
Established alignment across the teams by implementing a Jira board that accurately reflected the teams’ progress and product roadmap, allowing us to proceed with design and research planning
Ensured best usability practices by conducting a heuristic evaluation on previously completed design work and reviewing the user research to enhance the current design, before implementing new features
Project Deep Dive
Background
Mimik Studio is a healthtech start-up founded in collaboration with enFocus, a non-profit in South Bend, Indiana. Its founder, a speech-language pathologist, began building Mimik after recognizing the significant impact automation could have in clinical workflows.
In particular, she identified how SLPs constantly balance time and accuracy when manually transcribing patient sessions into the International Phonetic Alphabet (IPA). This phoneme-to-grapheme conversion is a core part of their daily work and often requires hours of focused effort after sessions, increasing fatigue and the risk of error. It also minimizes time spent provide treatment to patients
How Automation Can Reduce Cognitive Load
Research and literature indicated that transcription, not initial patient sessions or treatment, was the most time-intensive and error-prone part of an SLP’s workflow.
Based on our knowledge of Cognitive Load Theory (CLT) and vigilance decrement, automating transcription offered the greatest opportunity to reduce fatigue and errors. Our goal was to implement automation in ways that felt natural while maximizing clinician control so as not to interfere with clinical judgment.
Problem
” SLPs spend a large portion of their time devoted to manual transcription of sessions into the International Phonetic Alphabet. This takes time away from other clinical duties, including patient care and outreach. Automating this process would allow them to reallocate their time and potentially prevent transcription errors and abnormalities from going undetected due to fatigue.”
The team had moved past the discovery phase, completing user interviews with other SLPs and synthesizing the results. They had also finished a low-fidelity prototype of the most foundational pages for the application and started the development of the LLM that would power the transcription. My role was to continue working on the design, informed by the completed research.
Additional Opportunity: Product Management
As I began reviewing the user research and diving into the design, I recognized there was no true product management software in place. It was difficult to untangle where the different teams were in their processes, and we even struggled discussing which features belonged to which version of the project. I then also decided to take on a product management role to implement a Jira board that could lead the team to an Agile framework.
Before
Dead ends in designs marked as ready for development
Some things built without designs first
Disagreement on product features
Overall lack of visibility into progress across teams
After
Clear requirements outlined for each product version
Feature prioritization as a north star
Alignment across design, engineering, and business teams
Visible progress tracking across teams
Goals & Tasks
Product Management to Pave a Path Forward
I started by untangling the information architecture of the application and defining what success looked like for various future releases. I caught up with each team on what had been completed thus far and the roadmap ahead, and ensured that all of this information was reflected in Jira.
This allowed us to:
More easily discuss the product roadmap by having a north star to follow as a team
Strategically move forward with design, knowing what to prioritize for the MVP
Begin planning our research as the product proceeded for clinical trials
Design Process
Simplified Screen Designs
Screens simplified in compliance with an NDA
Patient List
Goal: Organize complex patient data, maximizing navigability
Design Focus: Clear hierarchy within the cards, flexible filter and search options, separating inactive from active patients
Patient Profile Creation
Goal: Design and utilize data entry fields that would be more usable for a mobile interface
Design Focus: Avoiding design patterns that would not be intuitive to the mobile interface or users with motor issues
Conducting Sessions
Goal: A no-fuss experience for SLPs to capture the patient's speech without interfering with their presence during sessions
Design Focus: Simple interactions and clear feedback on how and if the device is picking up the patient's speech, option to make edits once recording is paused
Session Reports
Goal: Surface analytics about the session for the SLP's review, another route to review and edit transcripts after processing, and space for patient notes
Design Focus: Quick, accurate, and scannable insights, and control for the SLP's
Outcomes
Ready for Clinical Trials
Delivered a complete, high-fidelity prototype with validated user flows and information architecture, ensuring the product was design-ready for usability testing and upcoming clinical trials.
Usability Study Planned
Collaborated with a researcher to design a System Usability Scale (SUS) study, establishing a clear framework for evaluating usability and collecting standardized metrics during clinical trials.
Jira Implemented
Introduced Jira as the team’s product management system, translating design work, engineering progress, and research priorities into an organized Agile workflow shared across teams.
Product Roadmap Created
Defined success criteria for future releases and aligned features to specific versions, creating a product roadmap that clarified scope, reduced ambiguity, and supported scalable development.