Ghana English Speech: 600 Hours of News Media Audio for ASR

A 600-hour speech dataset of Ghanaian English extracted from Ghanaian news media broadcasts, designed for training Automatic Speech Recognition models on West African accents. The dataset was created by the ghananlpcommunity and was last updated on March 6, 2026. It contains audio segments with verbatim transcriptions and duration metadata.

Use Cases

• Fine-tuning ASR models for West African English accents based on the described Ghanaian English audio.
• Training accent-specific speech recognition systems based on the 600 hours of news media broadcasts.
• Benchmarking ASR model performance on non-native English varieties based on the provided verbatim transcriptions.
• Studying phonetic and linguistic features of Ghanaian English based on the audio and corrected text columns.

Strengths

• Contains 600 hours of audio data, a substantial volume for model training.
• Provides verbatim transcriptions (corrected_text) aligned with audio segments.
• Audio is sampled at 24 kHz, a standard quality for speech processing.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic and source bias inherent to the news media collection.

Provenance

Source

ghananlpcommunity

Collection Method

Extracted from Ghanaian news media broadcasts.

Time Range

null

Freshness

Last updated 2026-03-06 23:07:19; freshness should be verified.

Geography

Ghana, West Africa