Sensory
Hearing loss
A partial or complete loss of hearing in one or both ears. It is classified as conductive, sensory, or central.
Living starter kit
Not yet expert-checkedUpdated Nov 13, 2026Start building on hearing loss
Hearing loss is one of the most builder-friendly areas on Unsolved: open audio datasets with listening-test results from people with hearing loss, and well-run challenges with baselines and code. The hard part is improving what people actually hear in real rooms and with music, not scores on speech tests designed for normal hearing.
Know hearing loss well? No expert has checked this kit yet.Check our work →Open datasets4
- Clarity CPC2 hearing-aid intelligibility data ↗
Speech processed by hearing-aid algorithms, with each listener's hearing profile and how many words they understood, from the Clarity project.
12.6 GB of files on the release page
AudioOpen downloadLicense not stated
- Clarity CEC3 real-room scenes ↗
Noisy scenes built with real measured room acoustics, for training hearing-aid speech enhancement, from the Clarity project.
12.5 GB of development and evaluation files
AudioOpen downloadLicense not stated
- Cadenza music data (first challenge) ↗
Music for training, validating and testing systems that personalize music for listeners with hearing loss, from the Cadenza project.
30.4 GB of files on the release page
AudioOpen downloadLicense not stated
- NHANES 2017–2018 hearing exams ↗
Hearing test results and ear examinations from the U.S. National Health and Nutrition Examination Survey.
ClinicalOpen downloadLicense not stated
Benchmarks and tasks3
- Clarity CPC2 intelligibility prediction ↗Predict how many words a listener with hearing loss will understand, from processed speech and their hearing profile. Audio
- Clarity CEC3 speech enhancement ↗Clean up noisy hearing-aid audio for listeners with hearing loss, in increasingly realistic rooms. Audio
- Cadenza music enhancement (first challenge) ↗Separate and remix music to suit a listener's hearing while keeping it sounding good. Audio
Open questions4
- 01
Can intelligibility prediction work for new listeners and new hearing-aid algorithms?
The challenge deliberately holds out listeners and systems to test exactly this.
AudioClinicalSource: CPC2 task and evaluation ↗
- 02
Can speech enhancement trained on simulated rooms work in real ones, with moving sounds?
Clarity's latest challenge adds real room recordings because training on simulations often fails in practice.
- 03
Can music be personalized for hearing loss without making it sound worse?
Hearing aids are tuned for speech and often handle music poorly.
- 04
Can a phone hearing test reliably pick out who needs a full hearing exam?
WHO's hearWHO app is a screener, not a diagnosis, and isn't meant for young children.
AudioClinicalSource: hearWHO hearing-check guidance ↗
Who's working on this4
Clarity project (Sheffield, Salford, Nottingham, Cardiff)
The consortium runs hearing-aid speech challenges with open data and baseline software.
Its open toolkit has ready-to-run challenge recipes, and the team asks for contributions.
The toolkit explicitly encourages open-source system and model contributions; this is a code invitation, not a claim that past challenge submission windows remain open.
Cadenza project (Salford, Sheffield, Leeds, Nottingham)
The consortium runs music-processing challenges for listeners with hearing loss.
Music tasks for hearing loss, a less crowded alternative to generic audio source separation.
No public invitation foundEvidence, date unknown ↗
WHO hearWHO programme
WHO offers a phone-based hearing screener and guidance on what its results mean.
Its guidance shows where a screening app's limits are.
No public invitation foundEvidence, Mar 20, 2025 ↗
NHANES hearing exams, CDC
The survey releases hearing thresholds and ear examinations from a national sample.
Documented hearing measurements for population-level research.
No public invitation foundEvidence, date unknown ↗
Not in the kit yet
- The datasets don't state licenses; check terms before commercial use, including rights to the music. The toolkit's MIT license covers code, not data.
- The Clarity and Cadenza challenges listed have closed, but their data and evaluation labels remain available.
- No newborn hearing screening or cochlear implant datasets yet.
- The U.S. survey includes people with and without hearing loss, and its hearing thresholds don't measure understanding speech in noise.
This kit keeps growing as new data, benchmarks and teams appear. Know something we missed? Tell us →
Curated by the Unsolved project team. Every link was checked on the date shown; licenses are as stated by each source.
Where the field stands
01
Early understanding
No strong gene target or drug candidate yet
02
Target found
At least one gene target with strong evidence
03
Drug candidates
Preclinical or phase 1 drug candidates
04
In trials
Drug candidates in phase 2 or later
05
Has approved drugs
At least one approved drug; not necessarily a cure
Derived from Open Targets drug and target data, not self-reported.
Healthy years lost (DALYs), 2023
53.1M
95% range 37.5M to 71.6M
Deaths, 2023
Not modeled
People living with it, 2023
1.7B
GBD cause: Age-related and other hearing loss (exact match)
Research attention
What there is to build on
Open Targets Platform ↗Gene targets with strong evidence
100+
of 4,633 associated genes
Top associated genes
GJB2OTOFTECTAMYO7AMYO6
Drugs and clinical candidates
13
Approved drugs
0
Identifiers
- Open Targets
- MONDO_0005365
- MeSH
- D034381
- ICD-10
- H90
Trial and paper counts are rough measures of attention, not of progress. This page is for research planning, not medical advice.