Extracting Targeted Training Data from ASR Models, and How to Mitigate It

04/18/2022
by   Ehsan Amid, et al.
0

Recent work has designed methods to demonstrate that model updates in ASR training can leak potentially sensitive attributes of the utterances used in computing the updates. In this work, we design the first method to demonstrate information leakage about training data from trained ASR models. We design Noise Masking, a fill-in-the-blank style method for extracting targeted parts of training data from trained ASR models. We demonstrate the success of Noise Masking by using it in four settings for extracting names from the LibriSpeech dataset used for training a SOTA Conformer model. In particular, we show that we are able to extract the correct names from masked training utterances with 11.8 the time. Further, we show that even in a setting that uses synthetic audio and partial transcripts from the test set, our method achieves 2.5 accuracy (47.7 augmentation method that we show when used in training along with MTR, provides comparable utility as the baseline, along with significantly mitigating extraction via Noise Masking across the four evaluated settings.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
07/17/2020

CTC-Segmentation of Large Corpora for German End-to-end Speech Recognition

Recent end-to-end Automatic Speech Recognition (ASR) systems demonstrate...
research
06/14/2021

SynthASR: Unlocking Synthetic Data for Speech Recognition

End-to-end (E2E) automatic speech recognition (ASR) models have recently...
research
03/25/2023

Auto-AVSR: Audio-Visual Speech Recognition with Automatic Labels

Audio-visual speech recognition has received a lot of attention due to i...
research
09/30/2021

SpliceOut: A Simple and Efficient Audio Augmentation Method

Time masking has become a de facto augmentation technique for speech and...
research
05/13/2022

Who Are We Talking About? Handling Person Names in Speech Translation

Recent work has shown that systems for speech translation (ST) – similar...
research
03/04/2021

Error-driven Fixed-Budget ASR Personalization for Accented Speakers

We consider the task of personalizing ASR models while being constrained...
research
08/10/2017

Location Name Extraction from Targeted Text Streams using Gazetteer-based Statistical Language Models

Extracting location names from informal and unstructured texts requires ...

Please sign up or login with your details

Forgot password? Click here to reset