FLORAS: 50-Language Benchmark for Long-Form Conversational Speech

FLORAS is a 50-language benchmark for long-form recognition and summarization of spoken language. It was created by espnet and last updated on November 29, 2024. The dataset aims to provide a realistic test environment for models by using raw, long-form conversational audio with one or many speakers.

Use Cases

• Benchmarking speech recognition models based on raw long-form conversational audio.
• Evaluating speech translation models on multilingual, multi-speaker content.
• Testing automatic summarization systems for spoken language across 50 languages.
• Training models to handle realistic audio scenarios that differ from read-speech benchmarks.

Strengths

• Covers 50 languages, providing a broad multilingual scope.
• Focuses on realistic long-form conversational audio, unlike typical read-speech benchmarks.
• Created by espnet, a known organization in speech processing.

Limitations

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

Provenance

Source

huggingface

Freshness

Last updated 2024-11-29 20:12:19; freshness should be verified.