5 papers
AnchorPrompt: Self-Distilled Soft Prompts for Robust Audio-Language Models
Pooneh Mousavi, Amir Ivry, Mirco Ravanelli +1
Large audio-language models (LALMs) are sensitive to input perturbations, such as noise, waveform corruption, and adversarial injections. We propose AnchorPrompt, an efficient adap…
Voice Agents under Acoustic Stress: From Signal Degradation to Interaction and Action
Amir Ivry, Kai-Wei Chang, Lin Zhang +2
Voice agents must complete users' tasks despite noise, reverberation, and competing speech. Evaluating agents' robustness therefore requires following how acoustic conditions affec…
Task-Aware Answer Preservation under Audio Compression for Large Audio Language Models
Amir Ivry
Large audio language models (LALMs) increasingly reason over long audio clips, motivating audio compression to reduce memory use and inference latency. However, audio compression c…
LALM-as-a-Judge: Benchmarking Large Audio-Language Models for Safety Evaluation in Multi-Turn Spoken Dialogues
Amir Ivry, Shinji Watanabe
Evaluation of socially unsafe content in spoken dialogues remains text-centric, missing prosody and transcription failures. We present LALM-as-a-Judge, which includes an open bench…
MAPSS: Manifold-based Assessment of Perceptual Source Separation
Amir Ivry, Samuele Cornell, Shinji Watanabe
Objective assessment of audio source-separation systems still mismatches subjective human perception, especially when interference from competing talkers and distortion of the targ…