alakxender/dhivehi-audios-ds2

Dhivehi Audio Dataset 2

A quality-filtered Dhivehi speech dataset with three subsets (bronze, silver, gold), each representing a progressively stricter quality threshold.

Subsets

Subset	MOS	CTC gc	WER	Train	Test
bronze	≥3.0	≥0.70	—	65,838	7,316
silver	≥3.0	≥0.70	≤0.30	43,010	4,779

License

MIT | | gold | ≥3.5 | ≥0.75 | ≤0.20 | 7,058 | 785 |

• MOS — Mean Opinion Score (1–4), subjective listening quality rating
• CTC gc — CTC forced-alignment geo-confidence (0–1), measures pronunciation accuracy
• WER — Word Error Rate from ASR transcription vs reference text (lower is better)

Each subset is a strict superset of the next: gold ⊂ silver ⊂ bronze.

Fields

Field	Type	Description
audio	Audio (16 kHz, mono)	Speech audio
sentence	string	Dhivehi text (numbers spelled out in Thaana — normalised for TTS)
raw_sentence	string	Original source text (may contain Arabic numerals and punctuation)
gender	string	Speaker gender (male / female)

Usage

from datasets import load_dataset

ds = load_dataset("alakxender/dhivehi-audios-ds2", "silver")
print(ds["train"][0])