metadata
title: README
emoji: 👀
colorFrom: indigo
colorTo: yellow
sdk: static
pinned: false
The End-to-end Speech Challenge (ESC) is a benchmark for assessing a single ASR system on a collection of eight different speech recognition datasets. The ESC datasets are sourced from different domains and cover a range of audio and text distributions (speaking styles, background noise, transcription requirements). ESC consists of:
- A Hugging Face dataset to easily download and use pre-prepared audio-text data
- Checkpoints and scripts to reproduce runs for the five official baseline systems
- A leaderboard for ranking systems according to overall performance on the benchmark
ESC was proposed in ESC: A Benchmark For Multi-Domain End-to-End Speech Recognition by ... For more information, please see the official submission on OpenReview.net.