Sign in to view source links and access this dataset
Description
MathBridge_Audio is a multimodal dataset curated by Abigail Pitcairn, funded by the NSF and the University of Southern Maine AIIR Lab. It expands the original MathBridge dataset by Kyudan, containing 1,000 rows of English text translated to LaTeX with context, paired with audio recordings of speakers reading the full text. The dataset was last updated on August 14, 2025.
Use Cases
Training text-to-speech models for mathematical notation based on the paired English text and audio recordings.
Developing speech recognition systems for technical educational content using the audio of spoken mathematical text.
Creating or evaluating multimodal educational tools that link spoken language with LaTeX formatting.
Studying the alignment between natural language descriptions and formal mathematical representations.
Strengths
Contains a specific subset of 1,000 rows from the original MathBridge dataset.
Adds a novel audio modality to the existing text-to-LaTeX translation data.
Includes contextual information before and after each translation, as described.
Limitations
Row count is unknown for the audio component specifically, which may limit suitability assessment.
Column-level documentation is absent; field semantics must be inferred after download.
Description metadata is limited; actual data quality and audio characteristics require manual inspection.
Provenance
Source
Expansion of the MathBridge dataset by Kyudan, curated by Abigail Pitcairn.
Collection Method
Subset of the original dataset with added audio files of speakers reading the text.
Freshness
Last updated 2025-08-14 23:54:53; freshness should be verified.
License details are listed as 'open-source' but require checking the full description on the dataset page for specific terms.