Chinese Courtroom Audio and Case Data for Gender Speech Analysis in Yunnan, 2018

1,248 drug-related criminal trial cases from basic-level courts in Yunnan Province, China, in 2018 form the final analytic sample. The dataset includes matched audio files, judicial decisions, and replication scripts for a study on gender variation in Mandarin proficiency. Dong Yu created this repository on Harvard Dataverse to support replication of the original research.

Use Cases

• Train or evaluate speech processing models based on real-world courtroom audio.
• Analyze correlations between speaker gender and machine-predicted Mandarin proficiency scores.
• Conduct socio-legal research on case outcomes and proceedings in Chinese criminal trials.
• Replicate statistical analyses on gender variation using the provided data-merging and analysis scripts.

Strengths

• Includes 1,248 matched case records with audio and judicial documents, providing a linked multimodal resource.
• Draws from a broader catalog of over 500,000 trial videos across China from 2013 to 2019, suggesting a rigorous sampling frame.
• Provides detailed documentation of the workflow, including case selection, audio processing, and analysis steps.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data is limited to one province (Yunnan) and one year (2018), which may limit generalizability.
• The repository notes platform changes and copyright constraints may limit full replication of the original data collection.

Provenance

Source

China Court Trial Online and China Judgments Online.

Collection Method

Video and judicial decision records were collected, filtered, and matched; audio was processed and machine-predicted proficiency scores were generated.

Time Range

Focuses on trials from 2018; source videos span 2013-2019.

Freshness

Last updated 2026-06-02 18:53:52; freshness should be verified.

Geography

Yunnan Province, China.