Datasetpublic

sulabhkatiyar/ne-tts-lus

NE-TTS Mizo (lus) Cleaned TTS dataset for Mizo (lus), a North East Indian language. Derived from the Vaani dataset with SNR filtering, LUFS normalization, and text cleaning. Stats Metric Value Total clips 11,766 Total hours 19.9h High SNR (>=20dB) 8,554 clips Medium SNR (15-20dB) 3,212 clips Sample rate 16kHz Audio format WAV, 16-bit PCM Schema Column Type Description audio Audio 16kHz WAV audio text string… See the full description on the dataset page: https://huggingface.co/datasets/sulabhkatiyar/ne-tts-lus.

sourceHugging Facecc-by-4.0updated 4mo agoView on Hugging Face

0likes163downloads

Dataset Card

NE-TTS Mizo (lus)

Cleaned TTS dataset for Mizo (lus), a North East Indian language. Derived from the Vaani dataset with SNR filtering, LUFS normalization, and text cleaning.

Stats

Metric	Value
Total clips	11,766
Total hours	19.9h
High SNR (>=20dB)	8,554 clips
Medium SNR (15-20dB)	3,212 clips
Sample rate	16kHz
Audio format	WAV, 16-bit PCM

Schema

Column	Type	Description
audio	Audio	16kHz WAV audio
text	string	Cleaned transcript (Latin script)
language	string	ISO 639-3 code (lus)
gender	string	Speaker gender
district	string	Recording district
speaker_group	string	{gender}_{district}
snr	float	Estimated SNR in dB
duration	float	Clip duration in seconds
snr_tier	string	"high" (>=20dB) or "medium" (15-20dB)

License

CC-BY-4.0 (inherited from Vaani dataset).