# To Reverse the Gradient or Not: An Empirical Comparison of Adversarial and Multi-task Learning in Speech Recognition

arXiv: https://arxiv.org/abs/1812.03483 · HTML: https://mlsanity.com/paper/1812.03483/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Dec 2018.

## Built on by (most cited first, 10 of 10)

- [Diverse and Aligned Audio-to-Video Generation via Text-to-Video Model Adaptation](https://mlsanity.com/paper/2309.16429/) (2023)
- [textless-lib: a Library for Textless Spoken Language Processing](https://mlsanity.com/paper/2202.07359/) (2022)
- [Are disentangled representations all you need to build speaker anonymization systems?](https://mlsanity.com/paper/2208.10497/) (2022)
- [Fair NLP Models with Differentially Private Text Encoders](https://mlsanity.com/paper/2205.06135/) (2022)
- [Robust Stuttering Detection via Multi-task and Adversarial Learning](https://mlsanity.com/paper/2204.01735/) (2022)
- [Cross-Corpora Spoken Language Identification with Domain Diversification and Generalization](https://mlsanity.com/paper/2302.05110/) (2023)
- [Adversarial multi-task underwater acoustic target recognition: towards robustness against various influential factors](https://mlsanity.com/paper/2411.02848/) (2024)
- [SITA: Learning Speaker-Invariant and Tone-Aware Speech Representations for Low-Resource Tonal Languages](https://mlsanity.com/paper/2601.09050/) (2026)
- [Privacy-Preserving Speech Representation Learning using Vector Quantization](https://mlsanity.com/paper/2203.09518/) (2022)
- [Hear No Evil: Towards Adversarial Robustness of Automatic Speech Recognition via Multi-Task Learning](https://mlsanity.com/paper/2204.02381/) (2022)

## Abstract (arXiv)

Transcribed datasets typically contain speaker identity for each instance in the data. We investigate two ways to incorporate this information during training: Multi-Task Learning and Adversarial Learning. In multi-task learning, the goal is speaker prediction; we expect a performance improvement with this joint training if the two tasks of speech recognition and speaker recognition share a common set of underlying features. In contrast, adversarial learning is a means to learn representations invariant to the speaker. We then expect better performance if this learnt invariance helps generalizing to new speakers. While the two approaches seem natural in the context of speech recognition, they are incompatible because they correspond to opposite gradients back-propagated to the model. In order to better understand the effect of these approaches in terms of error rates, we compare both strategies in controlled settings. Moreover, we explore the use of additional untranscribed data in a semi-supervised, adversarial learning manner to improve error rates. Our results show that deep models trained on big datasets already develop invariant representations to speakers without any auxiliary loss. When considering adversarial learning and multi-task learning, the impact on the acoustic model seems minor. However, models trained in a semi-supervised manner can improve error-rates.