Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.23313011Rights:
No information provided; contact the repository owner
The Indian Multimodal Cardiovascular Speech and Physiological (IMCSP) Database is a research resource integrating speech recordings, acoustic features, and clinical measurements to support the investigation of non-invasive speech biomarkers for cardiovascular health in the Indian population. The database was collected in collaboration with Sri Jayadeva Institute of Cardiovascular Sciences and Research, Bengaluru, and RNS Institute of Technology, Bengaluru. This Zenodo record represents the privacy-preserving public release of IMCSP. The speech recordings included in this release have undergone voice-masking and de-identification procedures intended to reduce speaker identifiability, and direct participant identifiers have been removed from the accompanying files and metadata. The original, unmodified patient voice recordings are not included in this public release. The current release contains 203 WAV recordings from 69 participants and 206 tabular records comprising 56 fields. Participants include healthy volunteers and individuals with cardiovascular conditions performing sustained vowel phonation, sequential number counting, and passage reading. The accompanying acoustic measurements include pitch, jitter, shimmer, harmonics-to-noise ratio, intensity, formant frequencies, Mel-frequency cepstral coefficients (MFCCs), RASTA-PLP coefficients, speaking rate, and pause characteristics. These are linked to available de-identified demographic and clinical parameters, including left ventricular ejection fraction, blood pressure, heart rate, and oxygen saturation. Missing measurements are retained explicitly, and three tabular records have no corresponding audio recording. The public IMCSP release is intended to support reproducible research in biomedical signal processing, multimodal data fusion, explainable machine learning, and speech-based cardiovascular assessment, including exploratory applications in remote monitoring and telemedicine. Relationship to the PhysioNet release The IMCSP database is also being submitted separately to PhysioNet. The PhysioNet submission contains the original, unmodified clinical voice recordings and is intended to serve as the canonical clinical-data release subject to PhysioNet's review and applicable access requirements. Accordingly, the two repository records represent distinct versions of the research resource: Zenodo: privacy-preserving public derivative containing voice-masked/de-identified speech recordings and associated research data. PhysioNet: original clinical dataset containing the unmodified speech recordings and associated data Following publication of the PhysioNet record, its DOI and citation will be added to this Zenodo record to provide a direct cross-reference between the public privacy-preserving derivative and the original clinical dataset. Data-use and privacy notice The voice transformations used in this release are intended to reduce the risk of speaker identification and should not be interpreted as a guarantee of complete anonymity. Users must not attempt to identify or re-identify participants or use the recordings for speaker identification, voice cloning, impersonation, or other purposes intended to determine participant identity. The dataset should be used only in accordance with applicable ethical requirements, institutional policies, participant consent conditions, and data-protection regulations.
Keyword(s):
Acoustic features; Biomedical signal processing; Cardiovascular disease; IMCSP; Indian Multimodal Cardiovascular Speech and Physiological Database.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)