Afrikaans Speech Corpus With Orthographic Transcriptions

Orthographically transcribed broadband speech for Afrikaans, one of South Africa's eleven official languages. Transcriptions are provided in XML format. The corpus was authored by Febe de Wet and was last updated in March 2026.

Use Cases

• Train automatic speech recognition models using the orthographic transcriptions paired with audio files.
• Benchmark speech synthesis systems by using the transcriptions as ground truth text for Afrikaans audio.
• Analyze phonetic or linguistic patterns in Afrikaans speech using the time-aligned XML transcriptions.
• Develop multilingual speech models for South Africa by combining this corpus with data from other official languages.

Strengths

• Includes orthographic transcriptions for each speech recording.
• Covers one of South Africa's eleven official languages.
• Data is provided in a structured XML format for transcriptions.

Limitations

• The exact number of audio recordings, hours of speech, and speaker count are unknown.
• Scope is limited to a single language (Afrikaans) within the broader corpus project.
• The 0.0 B size indicates a very small dataset, limiting its utility for training large models.

Provenance

Source

Febe de Wet via figshare.

Collection Method

Collection and orthographic transcription of broadband speech.

Freshness

Last updated March 2026.

Geography

South Africa.