# Speech Therapy

- **Event:** [The Future of Agentic AI in Healthcare - Abridge x Anthropic x Lightspeed](https://cerebralvalley.ai/e/abridge-hackathon)
- **When:** Sat, Jul 18 at 9:00 AM – 10:00 PM (PDT)
- **Where:** San Francisco, CA
- **Team:** [Kristiyan Vachev](https://cerebralvalley.ai/u/ChrisVachev)
- **GitHub:** https://github.com/KristiyanVachev/abridge-speech-therapy
- **Demo video:** https://youtu.be/2l4jnHoOa3w
- **Gallery:** https://cerebralvalley.ai/e/abridge-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/abridge-hackathon/hackathon/gallery/75

Pre-Check — situational speech profiling for speech-language pathologists
The clinic is a non-representative speaking situation: people who stutter often speak very differently with a clinician than on a phone call or ordering food — so SLPs assess a sample that doesn't reflect the disorder. Worse, standard ASR can't even capture the data: it erases, smooths, or garbles stuttered speech (we tested it — sound repetitions come back as the wrong words entirely).
Pre-Check runs a pre-visit session: a patient talks with a calm voice agent about their speech (the baseline sample), then takes a simulated phone call — ordering a pizza — that elicits real-situation speech. A voice interface built for this audience: tap-to-finish turns, no VAD cutoffs, it never hangs up on you. Recordings run through a two-engine pipeline: CrisperWhisper for verbatim words + timestamps (blocks detected from inter-word gaps, repetitions from tokens) fused with LLM-Dys for dysfluency type classification, anchored per turn. The SLP gets a one-page profile: how speech changed across situations, playable timestamped evidence for every detected event, and a side-by-side of what standard ASR heard vs. what was actually said.

---

Markdown version of https://cerebralvalley.ai/e/abridge-hackathon/hackathon/gallery/75. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt