datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
vl8-edge-noise-bank
vl8-edge noise bank
Curated noise bundle for vl8-edge-sdk's
synth.augment pipeline (issue #9). Repackaged from the
DEMAND corpus on Zenodo.
License: CC-BY-4.0
Source: DEMAND — Joachim Thiemann, Nobutaka Ito, Emmanuel Vincent
Format: 16 kHz mono 16-bit PCM, 6 channels (3 cafe + 3 street), ~40 MB
The SDK lazy-downloads noise-bank-v1.tar.gz from this repo on first
vl8 build --augment and verifies per-file SHA256 against the manifest at… See the full description on the dataset page: https://huggingface.co/datasets/ggix/vl8-edge-noise-bank.editflow-eval-datasets
EditFlow evaluation datasets
librispeech_pc_testset.tar.gz
realedit.tar.gz