Celebrity Speech Dataset from YouTube Videos

Over 1 million utterances from 6,112 celebrities were extracted from YouTube videos for speaker identification research. The dataset was created by Reverb and was last updated on the Hugging Face platform in August 2025.

Use Cases

• Train speaker embedding models using the 1 million+ utterances and associated speaker labels.
• Benchmark speaker identification systems on the 6,112 unique celebrity identities.
• Develop audio-visual speech models using the linked MP4 video files and audio tracks.
• Perform speaker diarization analysis on the segmented utterance files.

Strengths

• Contains over 1 million audio utterances.
• Covers a large number of 6,112 unique speakers.

Limitations

• The dataset is limited to celebrity speakers, which may not represent general population speech patterns.
• Data is sourced from YouTube, which may introduce variable audio quality and background noise.

Provenance

Source

YouTube videos uploaded by users.

Collection Method

Utterances were extracted from publicly available YouTube videos of celebrities.

Freshness

Last updated on the platform in August 2025.