Decoupling recognition and transcription in Mandarin ASR

Abstract

Much of the recent literature on automatic speech recognition (ASR) is taking an end-to-end approach. Unlike English where the writing system is closely related to sound, Chinese characters (Hanzi) represent meaning, not sound. We propose factoring audio -> Hanzi into two sub-tasks: (1) audio -> Pinyin and (2) Pinyin -> Hanzi, where Pinyin is a system of phonetic transcription of standard Chinese. Factoring the audio -> Hanzi task in this way achieves 3.9% CER (character error rate) on the Aishell-1 corpus, the best result reported on this dataset so far.

arXiv commentsubmitted to ASRU 2021

References in corpus (32)

• Applying Wav2vec2.0 to Speech Recognition in Various Low-resource Languages
• Unified Streaming and Non-streaming Two-pass End-to-end Model for Speech Recognition
• Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition
• Recent Developments on ESPnet Toolkit Boosted by Conformer
• FastCorrect: Fast Error Correction with Edit Alignment for Automatic Speech Recognition
• WeNet: Production oriented Streaming and Non-streaming End-to-End Speech Recognition Toolkit
• Multi-head Monotonic Chunkwise Attention For Online Speech Recognition
• A Comparison of Modeling Units in Sequence-to-Sequence Speech Recognition with the Transformer on Mandarin Chinese
• Efficient conformer-based speech recognition with linear attention
• Spike-Triggered Non-Autoregressive Transformer for End-to-End Speech Recognition
• A Further Study of Unsupervised Pre-training for Transformer Based Speech Recognition
• WNARS: WFST based Non-autoregressive Streaming End-to-End Speech Recognition
• Improved Conformer-based End-to-End Speech Recognition Using Neural Architecture Search
• Intermediate Loss Regularization for CTC-based Speech Recognition
• Darts-Conformer: Towards Efficient Gradient-Based Neural Architecture Search For End-to-End ASR
• Listen Attentively, and Spell Once: Whole Sentence Generation via a Non-Autoregressive Architecture for Low-Latency Speech Recognition
• Improving RNN transducer with normalized jointer network
• One In A Hundred: Select The Best Predicted Sequence from Numerous Candidates for Streaming Speech Recognition
• Transformer with Bidirectional Decoder for Speech Recognition
• Non-Autoregressive Transformer ASR with CTC-Enhanced Decoder Input
• Decoupling Pronunciation and Language for End-to-end Code-switching Automatic Speech Recognition
• Research on Modeling Units of Transformer Transducer for Mandarin Speech Recognition
• Gated Recurrent Fusion with Joint Training Framework for Robust End-to-End Speech Recognition
• Transformer-based Online Speech Recognition with Decoder-end Adaptive Computation Steps
• Transformer-based end-to-end speech recognition with residual Gaussian-based self-attention
• Boundary and Context Aware Training for CIF-based Non-Autoregressive End-to-end ASR
• Adversarial Joint Training with Self-Attention Mechanism for Robust End-to-End Speech Recognition
• Self-and-Mixed Attention Decoder with Deep Acoustic Structure for Transformer-based LVCSR
• Automatic recognition of suprasegmentals in speech
• Multi-QuartzNet: Multi-Resolution Convolution for Speech Recognition with Multi-Layer Feature Fusion
• Relaxing the Conditional Independence Assumption of CTC-based ASR by Conditioning on Intermediate Predictions
• Head-synchronous Decoding for Transformer-based Streaming ASR