NSFW Text-to-Speech Audio Dataset with 30 Characters and 1000+ Hours

A high-quality audio dataset designed for training and fine-tuning NSFW text-to-speech models. It includes over 1000 hours of audio from 30 characters, with annotations for emotion and sound effects. The dataset was created by DMC-ykfx33 and was last updated on Hugging Face in April 2026.

Use Cases

• Train NSFW text-to-speech models based on the high-quality audio samples.
• Fine-tune voice synthesis for specific character personas based on the 30 included characters.
• Develop emotion-aware TTS systems based on the 6000+ emotion labels (e.g., intimate, breathy, teasing).
• Incorporate non-speech sound effects into speech synthesis based on the 760+ sound labels (e.g., moan, sigh, laugh).

Strengths

• Contains over 1000 hours of audio data, providing substantial material for model training.
• Includes annotations for more than 6000 emotion instances and 760 sound effects.
• Features audio from 30 distinct characters, offering diversity in voice samples.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect source bias inherent to its collection method and purpose.

Provenance

Source

Hugging Face user DMC-ykfx33.

Collection Method

Method of gathering is not described in the provided metadata.

Time Range

Temporal coverage is not specified.

Freshness

Last updated 2026-04-13 07:41:28; freshness should be verified.

Geography

Spatial coverage is not specified.