Mohamed Khairy Arabic Speech Corpus with 430 Hours of Audio and Non-Verbal Transcriptions

A large-scale Arabic speech corpus containing approximately 430 hours of speech recordings and corresponding transcripts. The dataset is a first-of-its-kind resource that includes rich non-verbal transcriptions alongside the spoken text. It was created by author oddadmix and was last updated on the platform in June 2026.

Use Cases

• Train Arabic automatic speech recognition (ASR) models based on the 430 hours of transcribed speech.
• Develop models for non-verbal speech analysis based on the included rich non-verbal transcriptions.
• Build text-to-speech (TTS) systems for Arabic using the aligned audio and text data.
• Research paralinguistic features in Arabic speech based on the detailed non-verbal annotations.

Strengths

• Approximately 430 hours of Arabic speech data provides a substantial resource for model training.
• Includes pioneering rich non-verbal transcriptions, which are not commonly found in Arabic speech corpora.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface user oddadmix

Freshness

Last updated 2026-06-07 16:32:07; freshness should be verified.

Geography

Likely Arabic-speaking regions, but specific geography is not stated.