metadata
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.