FLEURS Farsi: Processed Persian Speech Dataset for Low-Resource Evaluation

FLEURS Farsi contains the Persian (Farsi, fa_ir) portion of the FLEURS speech dataset created by Google. This version has been processed into a Hugging Face datasets compatible format by MohammadGholizadeh and was last updated on June 5, 2025. The dataset is designed for evaluating speech recognition systems, particularly in low-resource scenarios.

Use Cases

• Train automatic speech recognition (ASR) models based on the described audio recordings.
• Benchmark speech recognition performance for Persian (Farsi) based on the dataset's design for low-resource evaluation.
• Conduct few-shot learning experiments on speech representations using the FLEURS framework.
• Evaluate multilingual speech models on the Persian language subset.

Strengths

• Dataset originates from Google's FLEURS project, a known benchmark for speech representation evaluation.
• Specifically processed for compatibility with the Hugging Face `datasets` library, which suggests ease of integration.

Limitations

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

Provenance

Source

Google FLEURS dataset

Collection Method

Processed subset of the original many-language speech dataset.

Freshness

Last updated 2025-06-05 06:44:29.

Geography

Likely contains data from Iran (fa_ir), given the language code.