Emo-com: Real Emotional Conversations for Supportive Speech

Emo-com is a speech dataset of genuine emotional conversations between people actively supporting each other. It was created by liva-ai and last updated on HuggingFace in May 2026. The data captures unscripted speech reflecting how people talk to close friends when sharing hardships, daily experiences, and jokes.

Use Cases

• Train emotion recognition models based on genuine emotional expression in speech.
• Study conversational dynamics and support strategies based on real mutual-support dialogues.
• Develop AI for empathetic dialogue systems based on the context of sharing hardships and jokes.

Strengths

• Contains genuine emotional expression, unlike acted or scripted speech common in other datasets.
• Captures conversations in the specific context of mutual support, as described.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

liva-ai

Collection Method

Recordings from support and community groups operated with licensed professionals.

Freshness

Last updated 2026-05-29 00:20:06; freshness should be verified.