Datasets:
Tasks:
Automatic Speech Recognition
Formats:
parquet
Languages:
Slovak
Size:
100K - 1M
ArXiv:
License:
Download data/train-00032-of-00123.parquet from NaiveNeuron/SloPalSpeech: direct link, hf CLI and curl.
- Browser
- Download file 493 MB
-
https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00032-of-00123.parquet
- Command line
-
hf download hf://datasets/NaiveNeuron/SloPalSpeech/data/train-00032-of-00123.parquet
-
curl -L -o train-00032-of-00123.parquet https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00032-of-00123.parquet
493 MB
- Xet hash:
- d77ec53c835e9c9a5dedc082fde4bd82338cb89f701857cc5d30c5666c457dba
- Size of remote file:
- 493 MB
- SHA256:
- 6fe2ece4a00300ebe3981ee94b2a0dc1179abde013835eece22b47538d181105
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.