AISHELL-2

The AISHELL-2 dataset is a Mandarin speech recognition benchmark used to evaluate automatic speech recognition (ASR) systems, particularly in the context of streaming recognition.

Papers using AISHELL-2 (12)

Streaming Speech Recognition with Decoder-Only Large Language Models and Latency Optimization2026End-to-end Speech Recognition with similar length speech and text2025EffectiveASR: A Single-Step Non-Autoregressive Mandarin Speech Recognition Architecture with High Accuracy and Inference Speed2024Consistent Training and Decoding For End-to-end Speech Recognition Using Lattice-free MMI2021 · 2 citesParaformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech Recognition2022 · 2 citesFew-Shot Speaker Identification Using Depthwise Separable Convolutional Network with Channel Attention2022 · 1 citesA context-aware knowledge transferring strategy for CTC-based ASR2022Improving Noisy Student Training on Non-target Domain Data for Automatic Speech Recognition2022ApproBiVT: Lead ASR Models to Generalize Better Using Approximated Bias-Variance Tradeoff Guided Early Stopping and Checkpoint Averaging2023Streaming Decoder-Only Automatic Speech Recognition with Discrete Speech Units: A Pilot Study2024Semi-supervised Learning for Code-Switching ASR with Large Language Model Filter2024Improving Mandarin End-to-End Speech Recognition with Word N-gram Language Model2022