Cantonese Audio and Transcripts from YouTube for ASR Training

Cantonese audio segments and creator-uploaded transcripts extracted from various YouTube channels. The dataset was created by OrcinusOrca and last updated on August 27, 2025. It is intended for training automatic speech recognition models.

Use Cases

• Train Cantonese speech recognition models based on audio segments and transcripts.
• Fine-tune ASR models on creator-provided transcripts, which are likely higher quality.
• Benchmark speech recognition performance on informal, real-world Cantonese audio from YouTube.

Strengths

• Includes creator-uploaded transcripts, which are likely higher quality than auto-generated ones.
• Audio is sourced from real-world YouTube channels, providing diverse content.

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 is unknown, which may limit suitability assessment.

Provenance

Source

YouTube channels.

Collection Method

Audio and available Cantonese subtitles were downloaded and processed.

Freshness

Last updated 2025-08-27 02:39:27; freshness should be verified.