Voice Recordings for Cognitive Impairment Classification, 320 Female Participants

Voice recordings from 320 female participants, including 105 with Alzheimer's disease, 92 with mild cognitive impairment, and 123 cognitively normal controls. The dataset was used in a study by Minsoo Kim, published on figshare in May 2026, to evaluate deep-learning models for audio-based diagnosis. The dataset is small, at 5.5 KB, and is stored in an XLS file format.

Use Cases

• Training audio classification models based on features like spectrograms, mel-spectrograms, and MFCCs.
• Benchmarking deep-learning architectures (e.g., 1D CNN, Audio Spectrogram Transformer, Wav2Vec 2.0) for medical speech analysis.
• Developing tools for the early detection of Alzheimer's disease and mild cognitive impairment based on voice characteristics.

Strengths

• Includes a clearly defined cohort of 320 female participants across three diagnostic categories.
• Associated research provides specific model performance metrics (e.g., Wav2Vec 2.0 accuracy of 0.83 for NC vs. AD classification).
• Released under a permissive CC-BY-4.0 license for open use.

Limitations

• Row count and specific column-level documentation are unknown, requiring manual inspection after download.
• The dataset is very small (5.5 KB), suggesting it may contain summary statistics or metadata rather than raw audio files.
• Data is limited to a single demographic (female participants), which may limit generalizability.

Provenance

Source

Derived from the AI Hub 'Cognitive Impairment Diagnosis Voice/Conversation' dataset.

Collection Method

Speech data were obtained and analyzed as part of a study comparing deep-learning models.

Freshness

Last updated 2026-05-21 17:41:06; freshness should be verified.