Improved Consistency Training For Semi-supervised Sequence-to-sequence ASR Via Speech Chain Reconstruction And Self-transcribing

Abstract

Consistency regularization has recently been applied to semi-supervised sequence-to-sequence (S2S) automatic speech recognition (ASR). This principle encourages an ASR model to output similar predictions for the same input speech with different perturbations. The existing paradigm of semi-supervised S2S ASR utilizes SpecAugment as data augmentation and requires a static teacher model to produce pseudo transcripts for untranscribed speech. However, this paradigm fails to take full advantage of consistency regularization. First, the masking operations of SpecAugment may damage the linguistic contents of the speech, thus influencing the quality of pseudo labels. Second, S2S ASR requires both input speech and prefix tokens to make the next prediction. The static prefix tokens made by the offline teacher model cannot match dynamic pseudo labels during consistency training. In this work, we propose an improved consistency training paradigm of semi-supervised S2S ASR. We utilize speech chain

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyqi2022improved

Related papers

• Speaker Consistency Loss And Step-wise Optimization For Semi-supervised Joint Training Of TTS And ASR Using Unpaired Text Data (2022)0.00
• Cycle-consistency Training For End-to-end Speech Recognition (2018)0.00
• Semi-supervised Sequence-to-sequence ASR Using Unpaired Speech And Text (2019)0.00
• Mcr-data2vec 2.0: Improving Self-supervised Speech Pre-training Via Model-level Consistency Regularization (2023)3.58
• Joint Training Of Speech Enhancement And Self-supervised Model For Noise-robust ASR (2022)0.00
• Sequence-to-sequence ASR Optimization Via Reinforcement Learning (2017)9.41
• Tokenchain: A Discrete Speech Chain Via Semantic Token Modeling (2025)0.00
• Listening While Speaking And Visualizing: Improving ASR Through Multimodal Chain (2019)4.52