Callyope

What does Callyope sell besides a scribe?

The vendor's headline reads "The Clinical AI Platform that understands psychiatry." On the clinician side it lists structured chart reviews built from complex records, with history, diagnosis, symptoms and treatments in one view, and an ambient scribe that generates structured clinical notes from a consultation. The vendor claims the chart reviews are "+50% more accurate than those generated by general-purpose LLMs." Document drafting for hospitalization reports and referral letters is marked "Coming soon."

The patient side is a separate product that a clinician prescribes. The patient records voice journals, answers validated symptom questionnaires at a frequency the clinician sets, and completes brief automated check-ins. The vendor says a speech and behavior foundation model sits underneath, trained on voice, language, clinical context, sleep and activity signals, and that "with 30 seconds of speech, Callyope can assess 10+ mental health symptoms with over 90% accuracy." The page cites no study for that figure. The product works in English and French, and the vendor says eight clinical trials are ongoing with leading institutions.

A French company with EUR 2.2M of disclosed funding

The legal entity is CALLYOPE, a société par actions simplifiée with its registered office in Neuilly-sur-Seine. Press coverage places the company at Station F in Paris and names Martin Denais as CEO, Xuan-Nga Cao as COO and Rachid Riad as CSO and CTO. The founding year is not stated. EUR 2.2M was announced in November 2023, led by 360 Capital and Bpifrance Digital Venture with No Label Ventures and angels, and Maddyness gave a valuation of EUR 6.55M. The site gives no count of clinics or clinicians using the product.

What has Callyope's speech research published?

Three PubMed papers cover the company's speech work, and all three have Callyope-affiliated authors. The closest to the voice-analysis claim is a 2024 JMIR study (PMID 39324329) that recorded speech from 865 healthy adults through a mobile app. The authors reported an AUC of 0.76 to 0.78 for depression, 0.77 for anxiety, 0.73 for insomnia and 0.68 for fatigue. A 2026 JMIR Mental Health study (PMID 42607290) applied LLM-based topic modeling to spontaneous speech in depression across four multilingual cohorts, and a 2025 Scientific Reports paper (PMID 41372338) covered Parkinson's disease detection from speech.

Does Callyope publish a price?

Callyope lists no public price.

The verdict

A psychiatric service in France that wants questionnaires and voice check-ins between consultations, with chart summaries and a scribe on the same platform, is the intended customer. For a US practice the missing pieces are a HIPAA attestation and a named EHR. The closest paper to the voice claim studied healthy adults rather than patients and was written by the vendor's own researchers.