SinTechSVS: Phoneme-Level Singing Technique Annotations

A phoneme-level manual annotation dataset for singing techniques, covering pitch and timbral dimensions. The dataset was produced as part of the SinTechSVS project by the NUS Sound and Music Computing Lab. It is associated with a paper published in IEEE/ACM TASLP 2024.

Use Cases

• Training singing voice synthesis models based on phoneme-level technique annotations.
• Analyzing the relationship between pitch technique and timbral technique in singing.
• Developing evaluation metrics for singing technique generation in voice synthesis systems.
• Studying the manual annotation process for singing performance features.

Strengths

• Manual annotation provides a high-quality ground truth.
• Covers two distinct technical dimensions: pitch and timbre.
• Associated with a peer-reviewed publication in IEEE/ACM TASLP 2024.

Limitations

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

Provenance

Source

NUS Sound and Music Computing Lab

Collection Method

Manual annotation

Freshness

Last updated 2026-06-22 16:35:14; freshness should be verified.