Gigaspeech2 Typhoon: 1,000 Thai Speech Samples for ASR Benchmarking

A metadata-only reference dataset containing 1,000 test samples for Thai speech recognition benchmarking. The dataset, created by typhoon-ai, provides audio IDs and human transcriptions derived from the Gigaspeech2 corpus, with the last update recorded on 2026-05-18. Each audio_id links to the original Gigaspeech2 dataset for audio file retrieval.

Use Cases

• Benchmark ASR model accuracy based on the provided human transcriptions
• Evaluate model performance on Thai speech using the standardized test set
• Link metadata to original audio files from the Gigaspeech2 corpus for model training or validation

Strengths

• Provides a standardized benchmark with 1,000 test samples specifically for Thai ASR
• Includes human transcriptions for ground-truth evaluation
• Offers direct links to the original Gigaspeech2 audio files via audio_id

Limitations

• Row count for the full dataset is unknown, which may limit suitability assessment
• Column-level documentation is absent; field semantics must be inferred after download

Provenance

Source

Derived from the Gigaspeech2 corpus.

Collection Method

Metadata and transcriptions extracted from a larger speech corpus.

Time Range

null

Freshness

Last updated 2026-05-18 04:22:03; freshness should be verified

Geography

null