Datasetpublic

somu9/iisc_mono_hindi_female

IISc Mono Hindi Female Studio-quality single-speaker Hindi female TTS dataset from the SYSPIN project by Indian Institute of Science (IISc), Bengaluru. Dataset Description Property Value Source IISc SYSPIN Project Speaker Single professional female voice artist (42 yrs, 21 yrs experience) Language Hindi (hi) Total Duration 54 hours 54 minutes 44 seconds Utterances 22,058 (train: 21,662 / test: 396 EVAL domain) Audio 48kHz, 24-bit, mono… See the full description on the dataset page: https://huggingface.co/datasets/somu9/iisc_mono_hindi_female.

sourceHugging Facecc-by-4.0updated 5mo agoView on Hugging Face

1likes51downloads

Dataset Card

IISc Mono Hindi Female

Studio-quality single-speaker Hindi female TTS dataset from the SYSPIN project by Indian Institute of Science (IISc), Bengaluru.

Dataset Description

Property	Value
Source	IISc SYSPIN Project
Speaker	Single professional female voice artist (42 yrs, 21 yrs experience)
Language	Hindi (hi)
Total Duration	54 hours 54 minutes 44 seconds
Utterances	22,058 (train: 21,662 / test: 396 EVAL domain)
Audio	48kHz, 24-bit, mono, embedded in parquet
Recording	Neumann TLM-103 microphone, professional studio, ~40dB SNR
Domains	Agriculture, Books, Education, Evaluation, Finance, General, Health, Others, Politics, Weather

Domain Distribution

Domain	Hours	Sentences	Description
BOOK	23:03:20	8,358	Books
OTHE	7:21:25	3,081	Others
GENE	6:00:43	2,540	General
EDUC	5:04:10	2,060	Education
WEAT	4:22:58	1,873	Weather
POLI	2:57:51	1,235	Politics
AGRI	1:48:18	848	Agriculture
HEAL	1:48:31	835	Health
FINA	1:46:45	832	Finance
EVAL	0:40:38	396	Evaluation (test set)

Fields

Column	Type	Description
`audio`	Audio (48kHz)	Speech waveform
`text`	string	Hindi transcription (Devanagari)
`domain`	string	Content domain (BOOK, GENE, etc.)
`speaker_id`	string	`hindi_female_spk001`
`language`	string	`hi`

Splits

—train: 21,662 utterances (all domains except EVAL)
—test: 396 utterances (EVAL domain — recommended by creators for TTS evaluation)

Usage

python

from datasets import load_dataset

ds = load_dataset("somu9/iisc_mono_hindi_female", split="train")

# Listen to first sample
print(ds[0]["text"])
audio = ds[0]["audio"]  # {"array": np.array, "sampling_rate": 48000}

Speaker Metadata

—Language: Hindi
—Gender: Female
—Age: 42
—Experience: 21 Years
—Languages known: Hindi, English, Tamil
—Mother tongue: Hindi

Recording Setup

—Microphone: Neumann TLM-103
—Environment: Professional studio
—Conditions: Studio quality at ~40dB SNR

License

This dataset is released under the CC-BY-4.0 license.

TTS data created under SYSPIN project by Indian Institute of Science, Bengaluru. The copyright in the TTS data belongs to Indian Institute of Science, Bengaluru.

Acknowledgments

We extend our heartfelt gratitude to the talented voice artist whose contributions were fundamental to this project's success. We are particularly grateful to the project of German Development Cooperation "FAIR Forward - AI for All" for their financial support in developing this TTS corpus, and Bhashini AI Solutions Private Limited for their financial support for part of the corpus beyond 44 hours for every voice artist in developing this TTS corpus.

Citation

bibtex

@misc{SYSPIN_S1.0_Corpus,
    Title = {SYSPIN_S1.0 Corpus - A TTS Corpus of 900+ hours in nine Indian Languages},
    Authors = {Abhayjeet Et al.},
    Year = {2025}
}

Contact

SPIRE Lab, EE Dept., IISc, Bengaluru Email: contact.syspin@iisc.ac.in