Skip to content
Fan Zeng (Frank)
Back to resources

GigaSpeechBench — Low-Resource Languages

This module benchmarks language and dialect coverage, including Southeast Asian languages and regional Arabic varieties.

Visit official source (opens in a new tab)

Research use

Test transfer under low-resource conditions and measure accuracy on fixed splits when comparing lightweight recognizers on the same hardware.

Limitations

These test sets represent general language coverage. Record language and dialect separately when reporting results.

License and access

Dataset license
CC BY 4.0 (opens in a new tab)
Access status
Direct access

Load the Low-Resource-Languages test configurations from the publisher’s public WebDataset release.

Additional sources