Unverified50% confidenceFactExact time
In all six model-dataset pairings, adding the target audio (text+audio) improved emotion recognition accuracy.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/10/2026
First Seen
Valid until: 2/8/2027
Sources
Related Entities
Related Claims
UnverifiedA new study proposes an interpretable singing emotion recognition framework that decodes model decisions using only a small set of singing-specific descriptors (covering prosody, spectrum, phonetics and vocal techniques), directly linking predictions to measurable acoustic phenomena.79% similarUnverified语音情感识别技术已能从语调、语速、停顿中识别细粒度情绪状态,准确率超过85%75% similarUnverifiedThe study uses SHAP to quantify each feature's contribution to emotion prediction and GRAD-CAM for qualitative evaluation of the learned representations of vocal techniques, the two complementing each other to form a complete explanation chain from input features to model decisions.75% similarUnverifiedThe design of learning merging coefficients per layer brings interpretability, as the differing degrees to which different layers rely on sentiment and ASR knowledge may reveal the hierarchical distribution of AD speech features.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1155374API
curl https://kongchang.com/api/v1/knowledge/claims/1155374MCP
get_claim(id=1155374)