CREMA-D emotional speech recordings

Source: Cao et al. (2014), CREMA-D: Crowd-sourced Emotional Multimodal Actors Dataset.
Original dataset: https://github.com/CheyneyComputerScience/CREMA-D

This 40-recording excerpt is made available under the Open Database License 1.0:
https://opendatacommons.org/licenses/odbl/1-0/
Individual recordings are licensed under the Database Contents License 1.0:
https://opendatacommons.org/licenses/dbcl/1-0/
Please retain this attribution and follow these licenses when reusing the data.

Catalog and recordings: https://oruk.ai/samples/emotional/catalog.json
The catalog lists each original source, audio hash, waveform, and saved model output.
Original WAV bytes are unchanged. Clips were matched by decoded-audio hash to
Oruk's training split and selected for clearly emotional delivery. These curated
training examples are demonstrations, not held-out evaluations or benchmarks.
Saved analysis is a real response from Oruk Resonance. Vocal labels describe the
recording and do not establish the speaker's feelings. Clicking Analyze this
sample makes a new request to the live model.