Is One Score Enough? Assessing Singing Quality of Songs with Temporal Score Curves
SoundMultimedia
Summary
The authors present SongSQA, a method to judge the quality of singing in full-length songs rather than just short clips. They address the challenge that evaluating long songs requires understanding how singing quality changes over time and how those changes affect the overall score, even though detailed segment labels are scarce. Their approach first predicts segment scores using a teacher model without manual labels, then combines these with learned features to rate the whole song while also showing quality changes over time. Their experiments show SongSQA improves accuracy compared to previous methods.
Authors
Yishan Lv, Jing Luo, Xinyu Yang, Zhizheng Wu
Abstract
Singing Quality Assessment (SQA) has become increasingly important for practical multimedia applications and Music AI systems, yet existing studies predominantly focus on short singing clips and remain insufficient for full-length songs. Unlike clip-level assessment, full-length song SQA requires modeling how singing quality varies across different audio segments and how these local variations influence the overall evaluation of vocal performance. Moreover, the scarcity of segment-level annotations makes effective supervision challenging, as directly assigning a single overall score label to every segment tends to treat different segment qualities as equivalent. To address these challenges, we propose SongSQA, a two-stage framework for full-length song SQA. In the first stage, a Segment Score Predictor is trained with pseudo labels generated by a pre-trained teacher model, enabling segment-level singing quality prediction without requiring manual segment annotations. In the second stage, a Song Quality Aggregator integrates segment features and predicted segment scores into unified segment embeddings, and employs a learnable song embedding together with self-attention to capture the connection between segment-level vocal performance and overall song quality. In this way, SongSQA dynamically aggregates critical quality cues across the song to produce a holistic quality prediction, while also generating a temporal segment-level quality curve. Experimental results demonstrate the effectiveness of SongSQA for full-length song SQA, achieving up to a 13.95% relative improvement in KTAU over the strongest baseline, while consistently improving other evaluation metrics across all datasets.