AISHELL-2: Transforming Mandarin ASR Research Into Industrial Scale

Abstract

AISHELL-1 is by far the largest open-source speech corpus available for Mandarin speech recognition research. It was released with a baseline system containing solid training and testing pipelines for Mandarin ASR. In AISHELL-2, 1000 hours of clean read-speech data from iOS is published, which is free for academic usage. On top of AISHELL-2 corpus, an improved recipe is developed and released, containing key components for industrial applications, such as Chinese word segmentation, flexible vocabulary expension and phone set transformation etc. Pipelines support various state-of-the-art techniques, such as time-delayed neural networks and Lattic-Free MMI objective funciton. In addition, we also release dev and test data from other channels(Android and Mic). For research community, we hope that AISHELL-2 corpus can be a solid resource for topics like transfer learning and robust ASR. For industry, we hope AISHELL-2 recipe can be a helpful reference for building meaningful industrial systems and products.

References in corpus (3)

• THCHS-30 : A Free Chinese Speech Corpus
• AISHELL-1: An Open-Source Mandarin Speech Corpus and A Speech Recognition Baseline
• Gated Recurrent Unit Based Acoustic Modeling with Future Context

Cited by in corpus (49)

• A Survey on Neural Speech Synthesis
• Adaptation Algorithms for Neural Network-Based Speech Recognition: An Overview
• LipSound2: Self-Supervised Pre-Training for Lip-to-Speech Reconstruction and Lip Reading
• ATCSpeech: a multilingual pilot-controller speech corpus from real Air Traffic Control environment
• Non-autoregressive Transformer-based End-to-end ASR using BERT
• Augmented Datasheets for Speech Datasets and Ethical Decision-Making
• Unsupervised pre-training for sequence to sequence speech recognition
• A Comparison of Label-Synchronous and Frame-Synchronous End-to-End Models for Speech Recognition
• Improving Mandarin End-to-End Speech Recognition with Word N-gram Language Model
• Integrating Lattice-Free MMI into End-to-End Speech Recognition
• Exploiting Pre-Trained ASR Models for Alzheimer's Disease Recognition Through Spontaneous Speech
• Universal ASR: Unifying Streaming and Non-Streaming ASR Using a Single Encoder-Decoder Model
• CN-Celeb: multi-genre speaker recognition
• Joint Echo Cancellation and Noise Suppression based on Cascaded Magnitude and Complex Mask Estimation
• Fast End-to-End Speech Recognition via Non-Autoregressive Models and Cross-Modal Knowledge Transferring from BERT
• CIF: Continuous Integrate-and-Fire for End-to-End Speech Recognition
• Improving Noise Robustness of an End-to-End Neural Model for Automatic Speech Recognition
• MetricNet: Towards Improved Modeling For Non-Intrusive Speech Quality Assessment
• Atss-Net: Target Speaker Separation via Attention-based Neural Network
• Cantonese Automatic Speech Recognition Using Transfer Learning from Mandarin
• Non-Autoregressive Transformer ASR with CTC-Enhanced Decoder Input
• Research on Modeling Units of Transformer Transducer for Mandarin Speech Recognition
• Multi-Encoder-Decoder Transformer for Code-Switching Speech Recognition
• Acoustic Word Embedding System for Code-Switching Query-by-example Spoken Term Detection
• Improved Robust ASR for Social Robots in Public Spaces
• Domain Aware Training for Far-field Small-footprint Keyword Spotting
• A Modularized Neural Network with Language-Specific Output Layers for Cross-lingual Voice Conversion
• A New Approach to Accent Recognition and Conversion for Mandarin Chinese
• Building Bilingual and Code-Switched Voice Conversion with Limited Training Data Using Embedding Consistency Loss
• The HCCL Speaker Verification System for Far-Field Speaker Verification Challenge
• Boundary and Context Aware Training for CIF-based Non-Autoregressive End-to-end ASR
• Learned Transferable Architectures Can Surpass Hand-Designed Architectures for Large Scale Speech Recognition
• HI-MIA : A Far-field Text-Dependent Speaker Verification Database and the Baselines
• Raw Waveform Encoder with Multi-Scale Globally Attentive Locally Recurrent Networks for End-to-End Speech Recognition
• Improving RNN Transducer With Target Speaker Extraction and Neural Uncertainty Estimation
• Training Wake Word Detection with Synthesized Speech Data on Confusion Words
• Latency-Controlled Neural Architecture Search for Streaming Speech Recognition
• Cascade RNN-Transducer: Syllable Based Streaming On-device Mandarin Speech Recognition with a Syllable-to-Character Converter
• Speaker-aware speech-transformer
• Extremely Low Footprint End-to-End ASR System for Smart Device
• CarneliNet: Neural Mixture Model for Automatic Speech Recognition
• CIF-based Collaborative Decoding for End-to-end Contextual Speech Recognition
• Unit selection synthesis based data augmentation for fixed phrase speaker verification
• Towards Robust Speaker Verification with Target Speaker Enhancement
• Integrating Knowledge into End-to-End Speech Recognition from External Text-Only Data
• End-to-End Multi-Look Keyword Spotting
• Exploring Voice Conversion based Data Augmentation in Text-Dependent Speaker Verification
• Parameterized Channel Normalization for Far-field Deep Speaker Verification
• Text-Only Domain Adaptation for End-to-End Speech Recognition through Down-Sampling Acoustic Representation