Nigerian Common Voice: 158 Hours of Speech Recordings with Transcriptions

158 hours of audio recordings with corresponding text transcriptions, curated by benjaminogbonna. The dataset includes metadata like accent and locale and was last updated on March 30, 2025. It was created to address a gap in speech and language datasets for African accents.

Use Cases

• Training automatic speech recognition (ASR) models based on the 158 hours of transcribed audio.
• Improving model accuracy for specific accents based on the included accent metadata.
• Benchmarking ASR performance on Nigerian English speech data.
• Developing voice-enabled applications for Nigerian English speakers.

Strengths

• Contains 158 hours of audio data, providing a substantial resource for model training.
• Includes accent and locale metadata, which can help improve model performance for specific demographics.
• Explicitly curated to address a known gap in datasets for African accents.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Data may reflect geographic or demographic bias inherent to its collection method on the Common Voice platform.

Provenance

Source

huggingface user benjaminogbonna, likely derived from the Mozilla Common Voice project.

Collection Method

Likely contains crowd-sourced audio recordings and transcriptions.

Freshness

Last updated 2025-03-30 16:49:15; freshness should be verified.

Geography

Nigeria, based on the title and description focus.