Patients practice between sessions, following a plan their therapist set. Every attempt goes back to the therapist, who decides what happens next.
The patient's week
In the room
You can see what patients practiced between appointments, and how each attempt went.
Walk into the appointment knowing what the patient practiced since the last visit.
Adjust the coming week's practice without building it from scratch.
Every report and therapy plan is a draft until the therapist reviews and signs it.
Outside the room
Each patient gets a plan from their clinician, with exercises matched to where they are in therapy, and works through it between visits with guidance at each step.
The therapist chooses each step. Patients see only what they should practice.
Every attempt is recorded and organized for the therapist to review and adjust.
Streaks and progress over time keep patients going, and the therapist sees the activity behind them.
Patients work through the plan their therapist set from their phone, between sessions, and the therapist can see how it is going.
The loop
The session recording and notes turn into a structured draft.
Inside the platform
The platform is built around a structured learning path, from identifying moments of disfluency through to carrying the work into real conversations.
Clinicians tag moments of disfluency frame by frame with a standard taxonomy, from core behaviors to secondary behaviors and tension level. The same tool builds our dataset and supports our research partnerships.
The clinician assembles each patient's path from established techniques, then unlocks each step as the patient is ready.
Patients rehearse difficult conversations with a virtual counterpart. The clinician sets the scenario and difficulty.
A short self-report is captured at intervals. The clinician can see how the patient describes their own experience over time.
UpSpeech Labs
We built our own annotation tool, and practicing speech-language pathologists use it to tag disfluencies frame by frame.
Foundations
Clinicians and engineers work side by side. The speech-language pathologists who use the platform with patients review the product decisions we make.
Co-development partner
AI infrastructure grant
Security and data
Clinics trust us with sensitive recordings. We treat that data the way a clinic would, and a therapist always has the final say on what the AI produces.
Every clinic's data is kept separate by organization. One organization can never see another's patients or recordings.
Data travels over TLS, and recordings and databases are encrypted while stored.
Our servers and file storage are in the European Union, and we handle personal data under the GDPR.
Recordings are served through short-lived, signed links, never from a public location.
Recordings are used to improve our models only where the patient has opted in. Identifying details are removed first, the recordings stay inside UpSpeech, and a patient can withdraw at any time.
We're working with a cohort of clinics and would like to hear from others working in speech therapy. Tell us about your practice and we'll be in touch.
