Research Foundation
Published Research
"EaseTalk: An LLM-Driven Speech Practice Tool for Real-Life Scenarios"
Our foundational research has been published in IEEE proceedings, establishing the technical and theoretical framework for AI-assisted speech practice.
View IEEE PublicationOur Mission
Improve communication confidence and quality of life through evidence-based, accessible AI technology.
Core Values
- Authentic communication over perfect fluency
- Connection and expression first
- Self-advocacy and empowerment
- Acceptance through visibility
Designed with those who matter the most:
Pure code is never enough. EaseTalk's architecture marries therapeutic rigour with lived experience at every milestone:
- Iterative prototype testing with student clinicians and people who stutter (PWS).
- Therapist-approved exercise catalogue aligned with evidence-based fluency-shaping principles.
- Language-and-tone audits to ensure emotional safety and inclusivity.
"The AI role-play felt surprisingly authentic and let me rehearse tough phone calls without fear."
— Beta Tester #001
Clinical Pilot in Practice
EaseTalk has been live-demoed and tested with early adopters at the South Infirmary Victoria University Hospital, Cork. Feedback sessions with both clinicians and service-users inform every update.
"I praise EaseTalk's ability to promote personal agency and its elegant mood-tracking features."
— Treacy Murphy, Senior Speech & Language Therapist
Understanding Individual Differences
Key Research Finding: Stuttering Variability is Highly Individual
Recent research confirms that fear responses, triggers, and therapeutic needs differ dramatically between users. This individual variability is fundamental to understanding effective treatment approaches.
Research Evidence:
- Fear structures vary significantly between individuals
- Triggers are highly personalized and context-dependent
- Time pressure affects stuttering severity differently for each person
- Therapeutic needs require individualized approaches
EaseTalk's Response:
- Personalization layers in scenario design
- Adaptive AI responses to individual patterns
- Customizable time pressure settings
- Individual fear structure mapping
Exposure Therapy & Fear Structures
Research demonstrates that exposure therapy effectiveness depends on understanding individual fear structures and triggers. This finding directly informs EaseTalk's personalized approach:
Research Insights:
- Fear structures must be individually mapped
- Exposure must be tailored to personal triggers
- Time pressure is a significant but variable factor
- One-size-fits-all approaches are less effective
EaseTalk Implementation:
- Individual fear assessment and mapping
- Personalized scenario difficulty progression
- Adjustable time pressure settings
- Trigger-specific practice environments
Source: PMC9947508 - Exposure Therapy and Individual Fear Structures
Personalized Scenario-Based Training
Building on research showing individual variability in stuttering, EaseTalk's scenario-based approach incorporates personalization layers to address each user's unique needs, triggers, and therapeutic goals.
Our personalized scenarios adapt to individual patterns in:
- Fear-specific situations: Job interviews, phone calls, social interactions
- Time pressure sensitivity: Adjustable pacing and response expectations
- Trigger environments: Customized contexts based on individual challenges
- Comfort progression: Gradual exposure tailored to personal readiness
- Communication goals: Aligned with individual therapeutic objectives
Research Considerations
Studies in this field have examined several aspects of scenario-based practice:
- Practice Environment: Safe, judgment-free spaces for rehearsal
- Repetition Opportunities: Multiple attempts at challenging scenarios
- Gradual Exposure: Progressive difficulty levels
- Feedback Mechanisms: Constructive guidance during practice
- Confidence Building: Familiarity through repetition
- Real-world Transfer: Application to daily situations
"Your current scenario-based approach is excellent, but it needs personalization layers to address the reality that fear responses, triggers, and therapeutic needs differ dramatically between users."
— Research insight from individual variability studies
Traditional Approaches
Conventional scenario practice often involves:
- Standardized scenarios for all users
- Limited personalization options
- Fixed time pressure settings
- One-size-fits-all therapeutic approaches
EaseTalk's Personalized Approach
Research-informed features include:
- Individual fear structure mapping
- Personalized trigger identification
- Adaptive time pressure controls
- Customized exposure progression
Our Journey at a Glance
• Born as a final-year Computer Science thesis tackling speaking anxiety in everyday life.
• Co-designed in weekly sprints with the Speech & Language Therapy Department at UCC and the stuttering community.
• Research demoed at ACM SmartComp 2025 ("EaseTalk: An LLM-Driven Speech Practice Tool for Real-Life Scenarios").
• Founder Marco — a proud person who stutters — awarded UCC Student Entrepreneur of the Year 2025 for EaseTalk's social-impact vision.
• First real-world pilot launched at the South Infirmary, bringing EaseTalk to its first clinical users.
Academic Publication
Our poster paper "EaseTalk: An LLM-Driven Speech Practice Tool for Real-Life Scenarios" was presented at ACM SmartComp 2025.
View details on the SmartComp 2025 forum
Our Methodology
Collaborative Development Approach
Clinical Partnership
Weekly collaboration with speech therapists at UCC to ensure clinical validity and therapeutic alignment
User-Centered Design
Iterative testing with people who stutter to prioritize real-world needs and experiences
Evidence-Based
Built on peer-reviewed research in speech therapy, exposure therapy, and AI-assisted learning
Development Process
Literature Review
Comprehensive analysis of existing research on stuttering, exposure therapy, and AI applications
Clinical Consultation
Regular feedback sessions with speech-language pathologists
Prototype Testing
Iterative testing with early users to refine features and usability
Clinical Pilot
Real-world testing in clinical settings to validate effectiveness
Ongoing & Future Research
Reducing Waiting Times
Evaluating hybrid therapy models where EaseTalk supplements clinician time and shortens referral-to-treatment gaps.
Longitudinal Outcomes
3-month study tracking sustained changes in confidence, communication comfort, and quality of life measures.
AI Feedback Efficacy
Measuring how real-time, supportive AI guidance contributes to skill development and self-efficacy.