---
license: cc-by-4.0
pretty_name: "VCTK Speaker embeddings (tarred)"
tags:
- audio
- tts
- speaker-embeddings
- VCTK
---

# VCTK Speaker Embeddings (tarred)

Items: 109

This dataset ships as a single tar at the repo root. Members preserve paths like
`VCTK/<id>/audio.mp3` and `VCTK/<id>/speaker_latent.safetensors`.

See loader.py for example loading.

Attribution:
  Contains audio and embeddings derived from the CSTR VCTK Corpus. Distributed under CC BY 4.0; attribution required.