Lameus/en_spontaneous_profanity

Dataset from this article: Multimodal prediction of profanity based on speech analysis includes "cleaned_cv.zip" which consists of "Common voice" dataset's records with profanity speech. It can be treated as train data. "collected_records.zip" is manually collected from YouTube videos with spontaneous speech with profanity. Annotation files have the names of files and transcribed speech. The dataset was used to test solution for real-time profanity prediction, solution is located here: github

Ethical restraint: records have profanity speech