Exploratory Evaluation of Speech Content Masking
This work addresses privacy concerns in speech processing by exploring content masking, but it is incremental as it defines a new problem space with preliminary evaluations.
The paper tackles the problem of protecting speech content privacy by introducing a toy problem for content masking, evaluating baseline techniques that modify phone codes, and finds that different masking strategies and locations impact automatic speech recognition and speaker verification tasks.
Most recent speech privacy efforts have focused on anonymizing acoustic speaker attributes but there has not been as much research into protecting information from speech content. We introduce a toy problem that explores an emerging type of privacy called "content masking" which conceals selected words and phrases in speech. In our efforts to define this problem space, we evaluate an introductory baseline masking technique based on modifying sequences of discrete phone representations (phone codes) produced from a pre-trained vector-quantized variational autoencoder (VQ-VAE) and re-synthesized using WaveRNN. We investigate three different masking locations and three types of masking strategies: noise substitution, word deletion, and phone sequence reversal. Our work attempts to characterize how masking affects two downstream tasks: automatic speech recognition (ASR) and automatic speaker verification (ASV). We observe how the different masks types and locations impact these downstream tasks and discuss how these issues may influence privacy goals.