7 papers
AccentCL: Robust Accent Classification with Incremental Expansion
Mu-Ruei Tseng, Waris Quamer, Ghady Nasrallah +1
Accent classifiers are typically trained with a fixed label inventory and cannot accommodate new accent categories as new data becomes available. Moreover, accented speech corpora…
VOSSA: Voiceprint Optimization for Streaming Speech Architectures
Mu-Ruei Tseng, Waris Quamer, Ghady Nasrallah +1
Real-time voice conversion (VC) systems commonly rely on pretrained speaker embeddings from automatic speaker verification (ASV) models. While effective for speaker discrimination,…
PHONOS: PHOnetic Neutralization for Online Streaming Applications
Waris Quamer, Mu-Ruei Tseng, Ghady Nasrallah +1
Speaker anonymization (SA) systems modify timbre while leaving regional or non-native accent cues intact, which is problematic because such cues can reveal a speaker's first-langua…
TVTSyn: Content-Synchronous Time-Varying Timbre for Streaming Voice Conversion and Anonymization
Waris Quamer, Mu-Ruei Tseng, Ghady Nasrallah +1
Real-time voice conversion and speaker anonymization require causal, low-latency synthesis without sacrificing intelligibility or naturalness. Current systems have a core represent…
DarkStream: real-time speech anonymization with low latency
Waris Quamer, Ricardo Gutierrez-Osuna
We propose DarkStream, a streaming speech synthesis model for real-time speaker anonymization. To improve content encoding under strict latency constraints, DarkStream combines a c…
Disentangling segmental and prosodic factors to non-native speech comprehensibility
Waris Quamer, Ricardo Gutierrez-Osuna
Current accent conversion (AC) systems do not disentangle the two main sources of non-native accent: segmental and prosodic characteristics. Being able to manipulate a non-native s…
End-to-end streaming model for low-latency speech anonymization
Waris Quamer, Ricardo Gutierrez-Osuna
Speaker anonymization aims to conceal cues to speaker identity while preserving linguistic content. Current machine learning based approaches require substantial computational reso…