---
license: cc-by-4.0
---

This is a dataset is a fork of [librispeech_asr](https://huggingface.co/datasets/librispeech_asr) that defines each original split (like train-clean-100) as a split (named `train.clean.100`, with dots instead of hyphens). This allows you to download each part separately.

This fork also reports a `path` for each sample accurately.