Penpal: 300K Synthetic Handwriting Lines in 13 Styles

A synthetic dataset of 300,000 lines of English text rendered as pen strokes. The data was generated by a Graves-style handwriting synthesis model primed with 13 different real-person handwriting styles. It was authored by 'breitburg' and last updated on Hugging Face in May 2026.

Use Cases

• Train handwriting generation models based on the pen stroke sequences.
• Benchmark handwriting recognition systems on synthetic, style-varied data.
• Study the characteristics of different handwriting styles based on the 13 provided exemplars.
• Develop data augmentation techniques for offline handwriting analysis using the stroke-level structure.

Strengths

• Contains 300,000 lines of synthetic handwriting data.
• Includes data generated from 13 distinct real-person handwriting styles.
• Provides stroke-level coordinate data for detailed analysis of writing dynamics.

Limitations

• Row count and total file size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data is synthetic, which may not fully capture the variability of genuine human handwriting.

Provenance

Source

Hugging Face (author: breitburg)

Collection Method

Synthetically generated by a Graves-style handwriting synthesis model.

Freshness

Last updated 2026-05-19 19:49:43; freshness should be verified.