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

# EARS Speaker Embeddings (tarred)

Items: 2568

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

See loader.py for example loading.

Attribution:
  Contains audio and embeddings derived from the EARS dataset. Distributed under CC BY-NC 4.0; attribution required; commercial use is not permitted.