NSFW TTS Dataset with 30 Speakers and Over 118 Hours of Audio

30 speakers contributed over 300,000 audio clips for text-to-speech synthesis, with individual speaker durations ranging from 5 to 118 hours. The dataset, titled 'Nsfw Tts Dataset 30Speakers', was created by author DMC-ykfx33 and hosted on Hugging Face. It was last updated on April 17, 2026.

Use Cases

• Train text-to-speech models based on the multi-speaker audio clips.
• Develop voice cloning systems based on the extensive per-speaker duration data.
• Benchmark speech synthesis quality based on the large number of audio samples.
• Research speaker adaptation and multi-speaker TTS based on the 30 distinct speaker identities.

Strengths

• Includes audio from 30 distinct speakers.
• Contains over 300,000 individual audio clips.
• Total duration exceeds 118 hours for the largest speaker contribution.
• Speaker-specific statistics are provided, showing varied sample counts and durations.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's title and platform tags suggest it contains Not Safe For Work (NSFW) content, which may restrict its use.
• Row count and file formats are unknown, which may limit suitability assessment.

Provenance

Source

Hugging Face, author DMC-ykfx33.

Collection Method

Likely collected or synthesized for text-to-speech applications.

Time Range

null

Freshness

Last updated 2026-04-17 06:53:37; freshness should be verified.

Geography

null