# Continuous Soft Pseudo-Labeling in ASR

arXiv: https://arxiv.org/abs/2211.06007 · HTML: https://mlsanity.com/paper/2211.06007/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in General knowledge distillation methods, one of 12 directions in Knowledge distillation.
- Builds on 22 of its 32 references that are in the map.
- First posted on arXiv 11 Nov 2022.

## Placed in

- Knowledge distillation / [General knowledge distillation methods](https://mlsanity.com/distill/general-knowledge-distillation-methods/)

## Builds on (most cited first, 22 of 22)

- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence](https://mlsanity.com/paper/2001.07685/) (2020)
- [MixMatch: A Holistic Approach to Semi-Supervised Learning](https://mlsanity.com/paper/1905.02249/) (2019)
- [SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition](https://mlsanity.com/paper/1904.08779/) (2019)
- [Self-Attention with Relative Position Representations](https://mlsanity.com/paper/1803.02155/) (2018)
- [Listen, Attend and Spell](https://mlsanity.com/paper/1508.01211/) (2015)
- [Libri-Light: A Benchmark for ASR with Limited or No Supervision](https://mlsanity.com/paper/1912.07875/) (2019)
- [ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring](https://mlsanity.com/paper/1911.09785/) (2019)
- [Pushing the Limits of Semi-Supervised Learning for Automatic Speech Recognition](https://mlsanity.com/paper/2010.10504/) (2020)
- [Improved Noisy Student Training for Automatic Speech Recognition](https://mlsanity.com/paper/2005.09629/) (2020)
- [Self-Training for End-to-End Speech Recognition](https://mlsanity.com/paper/1909.09116/) (2019)
- [End-to-end ASR: from Supervised to Semi-Supervised Learning with Modern Architectures](https://mlsanity.com/paper/1911.08460/) (2019)
- [Self-training and Pre-training are Complementary for Speech Recognition](https://mlsanity.com/paper/2010.11430/) (2020)
- [Iterative Pseudo-Labeling for Speech Recognition](https://mlsanity.com/paper/2005.09267/) (2020)
- [Rethinking Evaluation in ASR: Are Our Models Robust Enough?](https://mlsanity.com/paper/2010.11745/) (2020)
- [SlimIPL: Language-Model-Free Iterative Pseudo-Labeling](https://mlsanity.com/paper/2010.11524/) (2020)
- [CAPE: Encoding Relative Positions with Continuous Augmented Positional Embeddings](https://mlsanity.com/paper/2106.03143/) (2021)
- [Momentum Pseudo-Labeling for Semi-Supervised Speech Recognition](https://mlsanity.com/paper/2106.08922/) (2021)
- [Semi-supervised ASR by End-to-end Self-training](https://mlsanity.com/paper/2001.09128/) (2020)
- [Kaizen: Continuously improving teacher using Exponential Moving Average for semi-supervised speech recognition](https://mlsanity.com/paper/2106.07759/) (2021)
- [Semi-Supervised Speech Recognition via Local Prior Matching](https://mlsanity.com/paper/2002.10336/) (2020)
- [Advancing Momentum Pseudo-Labeling with Conformer and Initialization Strategy](https://mlsanity.com/paper/2110.04948/) (2021)

## Built on by (most cited first, 6 of 6)

- [How to Scale Your EMA](https://mlsanity.com/paper/2307.13813/) (2023)
- [Fast Streaming Transducer ASR Prototyping via Knowledge Distillation with Whisper](https://mlsanity.com/paper/2409.13499/) (2024)
- [Pseudo2Real: Task Arithmetic for Pseudo-Label Correction in Automatic Speech Recognition](https://mlsanity.com/paper/2510.08047/) (2025)
- [Enhancing Automatic Chord Recognition via Pseudo-Labeling and Knowledge Distillation](https://mlsanity.com/paper/2602.19778/) (2026)
- [The Exact Worst-Case Tail Probability under Bounded Kurtosis](https://mlsanity.com/paper/2607.05226/) (2026)
- [A Practical Recipe for Semi-Supervised Federated ASR: Online Pseudo-Labels with Server Update Stabilization](https://mlsanity.com/paper/2609.25471/) (2026)

## Abstract (arXiv)

Continuous pseudo-labeling (PL) algorithms such as slimIPL have recently emerged as a powerful strategy for semi-supervised learning in speech recognition. In contrast with earlier strategies that alternated between training a model and generating pseudo-labels (PLs) with it, here PLs are generated in end-to-end manner as training proceeds, improving training speed and the accuracy of the final model. PL shares a common theme with teacher-student models such as distillation in that a teacher model generates targets that need to be mimicked by the student model being trained. However, interestingly, PL strategies in general use hard-labels, whereas distillation uses the distribution over labels as the target to mimic. Inspired by distillation we expect that specifying the whole distribution (aka soft-labels) over sequences as the target for unlabeled data, instead of a single best pass pseudo-labeled transcript (hard-labels) should improve PL performance and convergence. Surprisingly and unexpectedly, we find that soft-labels targets can lead to training divergence, with the model collapsing to a degenerate token distribution per frame. We hypothesize that the reason this does not happen with hard-labels is that training loss on hard-labels imposes sequence-level consistency that keeps the model from collapsing to the degenerate solution. In this paper, we show several experiments that support this hypothesis, and experiment with several regularization approaches that can ameliorate the degenerate collapse when using soft-labels. These approaches can bring the accuracy of soft-labels closer to that of hard-labels, and while they are unable to outperform them yet, they serve as a useful framework for further improvements.