Short-Form Video Subtitle Translation Benchmark with 1,080 Aligned Rows

180 rights-cleared short scripted clips form a benchmark for subtitle translation and localization workflows. The package contains 540 timestamped source subtitle segments, 1,080 aligned translation rows, and 540 SRT files across English, Spanish, and Chinese (Simplified). Authored by imaz regi for the AI Translate Video project, this synthetic dataset was last updated in May 2026.

Use Cases

• Evaluate segment-level translation quality based on aligned text rows across three languages.
• Benchmark subtitle timing preservation workflows based on SRT-style timestamps and segment identifiers.
• Test language-pair packaging and localization QA for scenarios like product demos, tutorials, and ecommerce clips mentioned in the description.
• Support research on audiovisual translation using a self-authored, rights-cleared corpus free of third-party media.

Strengths

• Contains 1,080 aligned translation rows across three languages (English, Spanish, Chinese Simplified).
• Includes 540 ready-to-use SRT subtitle files and 180 distinct scripted video clips.
• All source text and translations were authored specifically for this release, ensuring rights clearance.
• Covers 10 practical scenario domains including product demos, tutorials, and educational explainers.

Limitations

• Row count for the primary data tables is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred from file schemas after download.
• The dataset is small in scale at 878.5 KB, representing a focused benchmark rather than a large-scale corpus.

Provenance

Source

AI Translate Video project

Collection Method

Self-authored synthetic benchmark; all text created specifically for this release.

Freshness

Last updated 2026-05-14 04:34:49; freshness should be verified.