Loading...
Loading...
Speech recognition, text-to-speech, speaker identification, music classification, audio event detection
2,533 datasets
Database 'Documents of Ancient Greek Music' (dDAGM) 1.1 is a collection of all notes attested in ancient Greek musical documents from the standard edition by Pöhlmann and West. It was created by Dr Tosca A.C. Lynch of the University of Oxford and includes documents marked as belonging to the Classical/Hellenistic (BC) and Imperial (AD) harmonic systems. The database uses custom Greek musical notation fonts designed by Stefan Hagel of the Austrian Academy of Sciences.
A collection of 32-channel impulse responses recorded in a university corridor using a moving loudspeaker and an Eigenmike spherical microphone array. The dataset includes responses at two distances (1m and 2m), nine elevation angles from -40 to 40 degrees, and 36 azimuth angles at 10-degree resolution. Researchers from Tampere University created this data for work on sound event localization and detection of overlapping sources.
U–Pb dating of zircons from middle Permian tuffs in the Canning Basin of Western Australia reveals a conflict with established spore-pollen zonation. The dataset, associated with a 2017 Australian Journal of Earth Sciences article, presents ages from chemical abrasion-isotope dilution thermal ionisation mass spectrometry (CA-IDTIMS) analysis. It includes an age of 267.04 ± 0.14 Ma from the Microbaculispora villosa Zone, which is 1.7 million years younger than tuffs from the Dulhuntyispora granulata Zone.
Brisbane City Council provides event information for the Brisbane Festival, an annual international arts festival held each September. The dataset is a transformed extract from the Trumba Calendar API, limited to the next 1,000 events and updated daily. Brisbane Festival attracts an audience of around one million people every year with a program of theatre, music, dance, circus, opera, and major public events.
Proceedings from the first international conference on 'Alternative Histories of Electronic Music' held in London in April 2016. The conference was part of an AHRC-funded project exploring the work of musician Hugh Davies, who documented 560 studios in 39 countries in his 1968 catalog. The project was led by Dr James Mooney at the University of Leeds.
Nineteen sediment cores were collected from five salt marshes on Cape Cod, Massachusetts, by the United States Geological Survey between 2015 and 2016. The cores, up to 168 cm long, provide measurements of dry bulk density, carbon content, land surface elevation, and radionuclide activity. The data represent a chronosequence of tidal marsh restoration occurring between 2001 and 2010.
A single-case proof-of-concept study evaluates a personalized automatic speech recognition system for a patient with severe dysarthria and global aphasia. Davide Mulfari authored the report, which was last updated on June 3, 2026. The study compares the accuracy of a speaker-dependent VIVOCA system against rehabilitation professionals.
A 2017 study by Mory et al. presents U–Pb zircon dating and palynological data from the Canning Basin, Western Australia. The dataset highlights an apparent age conflict of 1.7 million years between tuffs in non-marine and marginal-marine facies from the Roadian–Wordian stages. It was published in the Australian Journal of Earth Sciences and is hosted by the Australian Ocean Data Network.
351,469,333 listening events from 55,190 users and 3,471,884 tracks, extended with acoustic features and country-level cultural data. Eva Zangerle of Universität Innsbruck created this dataset by augmenting the LFM-1b dataset with features from the Spotify API and cultural dimensions from Hofstede and the World Happiness Report. The dataset is described in a 2020 paper in the Transactions of the International Society for Music Information Retrieval.
Twenty synthetic audio soundscapes, each three minutes long, were created for studying the estimation of strong labels using crowdsourcing. The dataset includes reference annotations, crowdsourced annotation outcomes, and weak labels for each 10-second segment. It was created by researchers at Tampere University for a 2021 IEEE WASPAA paper.
A collection of 20 synthetic audio files, each 3 minutes long, created for studying the estimation of strong labels from crowdsourced annotations. The dataset includes reference annotations, estimated strong labels, and weak labels for each 10-second segment. It was created by researchers from Tampere University for a 2021 IEEE WASPAA workshop paper.
Questionnaire response data retrieved from participants at the Eurosonic Noorderslag 2023 conference. The dataset explores the perspectives of music industry professionals on music streaming services and recommender systems. It was created by researchers including Karlijn Dinnissen and is associated with a 2023 ACM conference paper.
Sediment cores from 11 locations across four micro-tidal salt marshes on the south shore of Cape Cod, Massachusetts, collected in 2013 and 2014. The dataset reconstructs vertical accretion and carbon burial rates over the past century using lead-210 and cesium-137 age models. It was created by Meagan J Eagle of the United States Geological Survey.
A multimodal dataset of 70,000 samples constructed by pairing handwritten digit images with spoken digit audio clips. The handwritten data is sourced from the MNIST database, and the spoken data is extracted from the Google Speech Commands dataset, with audio pre-processed into Mel Frequency Cepstral Coefficients. The dataset was created by Lyes Khacef and colleagues for research in multimodal fusion.
WFP’s Automated Disaster Analysis and Mapping (ADAM) system collected this dataset on a Category 1 storm from October 7-9, 2025. The data covers impacts in Antigua and Barbuda, Montserrat, Guadeloupe, Dominica, Saint Kitts and Nevis, Anguilla, Martinique, and the British Virgin Islands, with the storm center located near latitude 17.3, longitude -60.6. It was last updated on May 21, 2026, and is available in SHP and CSV formats.
A 7.2 KB dataset provides a multi-scenario evaluation harness for testing predictive signal damping models across three distinct high-fidelity environments. It includes a data file and a production-ready Python verification script for automated regression testing. The dataset was authored by Jamie Davis and last updated on May 28, 2026.
U–Pb dating of zircons from thin middle Permian tuffs in the Canning Basin of Western Australia reveals a conflict with established spore-pollen zonation. The youngest tuffs within the Microbaculispora villosa Zone yielded an age of 267.04 ± 0.14 Ma, which is 1.7 million years younger than tuffs associated with the Dulhuntyispora granulata Zone elsewhere in the basin. This dataset, associated with a 2017 journal article, presents the geochronological data underpinning this apparent conflict.
Eva Zangerle and colleagues from Universität Innsbruck created this dataset for the 2019 ISMIR conference. It combines audio features extracted from the Million Song Dataset with Billboard Hot 100 chart performance data. The dataset covers 515,576 songs representative of western commercial music released between 1922 and 2011.
49 real-life audio recordings totaling 189 minutes and 52 seconds, sourced from the TUT Acoustic Scenes 2016 dataset. The data includes soft labels representing estimated strong labels derived from crowdsourced annotations via Amazon Mechanical Turk. The dataset was created by Irene Martín-Morató of Tampere University for studying label estimation from crowdsourcing.
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.