Speech Gap
A concrete, already-patented measurement technology being brought into the NIPG / Semmelweis ecosystem. It is the specific instrument around which the near-term clinical collaboration is organised.
Public description
Speech Gap is a promising longitudinal measurement direction for precision psychiatry. Its intended value is not stand-alone diagnosis but repeated, low-burden measurement within the same person. Weekly short speech samples may reveal changes relative to the person’s own baseline and may support exploratory estimation or prediction of clinical trajectories. This use remains to be tested prospectively against repeated clinical reference assessments.
This is the vetted public wording. See Longitudinal within-person measurement for why the within-person framing is the scientific core rather than a hedge.
Demonstrated sensitivity
Speech Gap can distinguish people suffering from psychiatric problems from typical subjects in diverse cases; its sensitivity has been demonstrated in Hungarian and in English. The cross-language validation is documented in Kálmán et al., Temporal Speech Parameters Detect Mild Cognitive Impairment in Different Languages: Validation and Comparison of the Speech-GAP Test in English and Hungarian (DOI).
Prof. Dr. János Kálmán, a prominent Hungarian psychiatrist, was honoured by Gedeon Richter Plc with the Richter Merit Medal for his development of the Speech-GAP Test®.
Safeguard. Demonstrated sensitivity in a discrimination task is not the same as validated stand-alone diagnostic use, and not the same as validated prognostic use in therapy monitoring — which is the new and still-untested question.
Intellectual property
| Item | Detail |
|---|---|
| Patent family | WO2020128542A1 (international), US12161481B2 (granted US) |
| Title | Automatic detection of neurocognitive impairment based on a speech sample |
| Priority | 18 December 2018 |
| Filed | 16 December 2019 |
| Granted / published | 10 December 2024 |
| Named inventors | Gábor Gosztolya, Ildikó Hoffmann, János Kálmán, Magdolna Pákáski, László Tóth, Veronika Vincze |
| Original assignee | University of Szeged |
| Transfer | 2025 partial transfer to SZTE TTC Zrt. |
Records: US grant · granted patent PDF · international publication
Safeguard. Licensing terms must not be inferred or stated beyond the assignee record above. WO2020128542A1 and US12161481B2 belong to the same family, not two separate inventions. See Editorial safeguards.
Origin partner
Dr. Gábor Gosztolya — University of Szeged and HUN-REN Hungarian Research Centre for Linguistics. A leading expert bridging the “speech gap” between resource-rich languages such as English and less-resourced languages such as Hungarian.
Planned integration into the avatar-therapy project
Exploratory, non-comparative, nested in Semmelweis / VIKOTE’s funded VSMT project:
- 12-week therapy period, with weekly short speech measurements.
- Three or four in-depth BPD reference assessments, corresponding to an interval of roughly six or four weeks between assessments when baseline and endpoint are included.
- Analyses target within-person trends and the prediction of later reference scores.
- No separate untreated control group is intrinsic to, or needed for, this observational question.
- Planned safeguards: ClinicalTrials.gov registration and a PCI Registered Report. Registration, ethics approval, and Stage 1 in-principle acceptance are separate, not-yet-completed processes.
Known limitation and planned NIPG extension
The current software treats respiratory pauses as silence. This is a real limitation in BPD, where breathing changes — including possible heavy or rapid breathing — may themselves carry relevant information. The planned extension distinguishes silence, inhalation, exhalation, filled pauses, speech, and uncertain intervals, then tests whether respiratory features are reliable and clinically informative. This is a natural NIPG contribution in signal processing and feature engineering, layered on top of the Szeged-origin IP. → Respiratory signal extension
Design principle for data collection
The trend-oriented concept also dictates how sessions should be recorded: repeatable prompts, timestamps, task and protocol identifiers, recording quality, respiratory-event handling, and links to clinical reference dates. Together these turn short sessions into a longitudinal research asset rather than a pile of audio files. → Data governance and annotation
Related pages
Gábor Gosztolya · Semmelweis / VIKOTE · NIPG · University of Szeged · Longitudinal within-person measurement · Temporal speech parameters · Respiratory signal extension · PCI Registered Reports