AI-Generated Therapeutic Conversations

This dataset provides therapist-client onversations in four languages (english, spanish, portuguese, and french) where it can help to identify speakers, language, arousal, and therapeutic_context. Each utterance is labeled for speaker, language, sentiment, valence/arousal, and therapeutic_context—useful for speaker ID, language ID, and emotion/context modeling. Labels speaker_label: int language: en | es | pt | fr sentiment_label: POSITIVE=0, NEGATIVE=1, NEUTRAL=2, MIXED=3 valence: POSITIVE=0.7, NEGATIVE=-0.7, NEUTRAL=0.0, MIXED=-0.2 arousal: float Note: Annotations were generated with claude-3-5-haiku-20241022.