Lwazi Afrikaans TTS Corpus: Phonetically Balanced Speech

Phonetically balanced sentences from reference texts were recorded in a studio environment. The dataset contains orthographic transcriptions and phonemically aligned transcriptions in TextGrid format, paired with 16 KHz, 16-bit WAV audio files. This resource is designed for speech synthesis and natural language processing research.

Use Cases

• Training text-to-speech models based on studio-quality audio and aligned phonemic transcriptions.
• Phonetic analysis of Afrikaans using phonemically aligned TextGrid files.
• Developing speech recognition systems using orthographic transcriptions paired with high-quality audio.
• Studying phoneme coverage in a controlled, phonetically balanced corpus.

Strengths

• Includes phonemically aligned transcriptions (TextGrid files) for detailed phonetic analysis.
• Audio is recorded in studio quality at 16 KHz and 16 bits per sample.
• Released under a permissive CC-BY-4.0 license for reuse and modification.
• Sentences are phonetically balanced to ensure adequate phone coverage.

Limitations

• Key metadata such as row count, dataset size, and specific column names are not provided.
• The listed last update date of 2026-03-24 is a future date, indicating a potential metadata error.
• The organization of the author is unknown, limiting provenance details.

Provenance

Source

Etienne Barnard

Collection Method

Studio recordings of phonetically balanced sentences chosen from reference texts.

Time Range

null

Freshness

2026-03 24

Geography

null