mlsanityData to 14 Oct 2026

A Method to Reveal Speaker Identity in Distributed ASR Training, and How to Counter It

arXiv 2104.07815 ↗First posted 15 Apr 2021cs.CL cs.CR cs.LGcited by 8

Not on the map. Why?

Abstract (arXiv)

End-to-end Automatic Speech Recognition (ASR) models are commonly trained over spoken utterances using optimization methods like Stochastic Gradient Descent (SGD). In distributed settings like Federated Learning, model training requires transmission of gradients over a network. In this work, we design the first method for revealing the identity of the speaker of a training utterance with access only to a gradient. We propose Hessian-Free Gradients Matching, an input reconstruction technique that operates without second derivatives of the loss function (required in prior works), which can be expensive to compute. We show the effectiveness of our method using the DeepSpeech model architecture, demonstrating that it is possible to reveal the speaker's identity with 34% top-1 accuracy (51% top-5 accuracy) on the LibriSpeech dataset. Further, we study the effect of two well-known techniques, Differentially Private SGD and Dropout, on the success of our method. We show that a dropout rate of 0.2 can reduce the speaker identity accuracy to 0% top-1 (0.5% top-5).

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 26 of its 57 references that are in the map. First posted on arXiv 15 Apr 2021.

Builds on

26 of 57 references.

References · 26

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  2. 2016Deep Learning with Differential Privacycited by 2,728
  3. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
  4. 2019Deep Leakage from Gradientscited by 1,021
  5. 2018VoxCeleb2: Deep Speaker Recognitioncited by 875
  6. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 864
  7. 2019Towards Federated Learning at Scale: System Designcited by 832
  8. 2020Inverting Gradients -- How easy is it to break privacy in federated learning?cited by 712
  9. 2018Federated Learning for Mobile Keyboard Predictioncited by 572
  10. 2015Listen, Attend and Spellcited by 543
  11. 2018Exploiting Unintended Feature Leakage in Collaborative Learningcited by 520
  12. 2015Attention-Based Models for Speech Recognitioncited by 401
14 more
  1. 2020iDLG: Improved Deep Leakage from Gradientscited by 382
  2. 2014Deep Speech: Scaling up end-to-end speech recognitioncited by 354
  3. 2018Audio Adversarial Examples: Targeted Attacks on Speech-to-Textcited by 274
  4. 2014Differentially Private Empirical Risk Minimization: Efficient Algorithms and Tight Error Boundscited by 107
  5. 2018Federated Learning for Keyword Spottingcited by 95
  6. 2020InstaHide: Instance-hiding Schemes for Private Distributed Learningcited by 90
  7. 2014End-to-end Continuous Speech Recognition using Attention-based Recurrent NN: First Resultscited by 68
  8. 2020A Framework for Evaluating Gradient Leakage Attacks in Federated Learningcited by 64
  9. 2017Deep Speaker: an End-to-End Neural Speaker Embedding Systemcited by 58
  10. 2020Training Speech Recognition Models with Federated Learning: A Quality/Cost Frameworkcited by 46
  11. 2019Gradientless Descent: High-Dimensional Zeroth-Order Optimizationcited by 43
  12. 2020Training Keyword Spotting Models on Non-IID Data with Federated Learningcited by 36
  13. 2020Understanding Unintended Memorization in Federated Learningcited by 36
  14. 2020TextHide: Tackling Data Privacy in Language Understanding Taskscited by 33

Built on by

Cited by year

Citing papers per year: 2021: 2; 2022: 2; 2023: 3; 2024: 1.

8 papers.

  1. 2023GPT-FL: Generative Pre-trained Model-Assisted Federated Learningcited by 24
  2. 2021Revealing and Protecting Labels in Distributed Trainingcited by 17
  3. 2022Detecting Unintended Memorization in Language-Model-Fused ASRcited by 10
  4. 2021Jointly Learning from Decentralized (Federated) and Centralized Data to Mitigate Distribution Shiftcited by 7
  5. 2022Extracting Targeted Training Data from ASR Models, and How to Mitigate Itcited by 7
  6. 2023Unintended Memorization in Large ASR Models, and How to Mitigate Itcited by 5
  7. 2023Speech Privacy Leakage from Shared Gradients in Distributed Learningcited by 1
  8. 2024GDBR: Label Recovery Attack Against Partial Gradient Encryption in Federated Learningcited by 0