Research

Advancing Fluency Science Together

Fluens Health is more than a clinical tool — it is a research-driven initiative dedicated to improving how speech disfluencies are detected, classified, and supported in real time.
We actively collaborate with clinics, private therapists, universities, and research institutions to push the boundaries of speech technology in multilingual and clinical contexts.

Our Research Focus

1. Stutter Detection & Classification

We develop models that identify fluent, clonic, and tonic segments from streaming speech. Our work includes:

  • frame-level and event-level detection

  • synthetic stuttering generation for data augmentation

  • multilingual labeling consistency

  • generalization across accents and recording conditions

2. Stutter-Aware Speech Recognition

We explore speech recognition systems that remain accurate under disfluency pressure. This includes:

  • adapting ASR models (Whisper, CrisperWhisper an others) for stuttered speech

  • integrating stutter predictions into decoding

  • improving robustness for clinical settings and home practice

  • multilingual stutter-aware language modeling

3. Clinical-Grade Fluency Metrics

We research ways to turn real-time predictions into meaningful clinical indicators:

  • frequency and duration of disfluencies

  • progress tracking across sessions

4. Extending to Dysarthria

Our long-term roadmap expands beyond stuttering.
We are developing early prototypes for:

  • dysarthria detection and severity estimation

  • intelligibility scoring

  • Dysarthric Speech Recognition

  • personalized speech adaptation tools

Why Collaborate With Us

We welcome partnerships with clinics, private therapists, universities, and NGOs who want to advance clinical speech technology.
Together, we can collaborate on:

  • Data collection (only with explicit consent and strict GDPR safeguards)

  • Model evaluation across diverse speakers and languages

  • Joint publications and conference submissions

  • Pilot studies in therapy and educational settings

  • Designing new workflows for digital fluency support

We offer participating partners early access to tools, co-authorship opportunities, and co-branded visibility.

Let’s Build the Future of Fluency Research

If you’re interested in joint research, dataset collaboration, or evaluating our models in your institution, we would be delighted to connect.

Reach out and let’s explore how we can work together.

Get in touch