WavLM

WavLM is a dataset/benchmark used to evaluate the performance of language model-driven losses in alleviating phoneme hallucinations in low-bitrate neural speech coding by comparing decoded utterances against BERT representations of ground-truth transcriptions.

Papers using WavLM (6)

LLaSE-G1: Incentivizing Generalization Capability for LLaMA-based Speech Enhancement2025 · 4 citesFrom Hallucination to Articulation: Language Model-Driven Losses for Ultra Low-Bitrate Neural Speech Coding2026Distillation-based Layer Dropping (DLD): Effective End-to-end Framework for Dynamic Speech Networks2026Parameter Efficient Transfer Learning for Various Speech Processing Tasks2022Benchmarking Children's ASR with Supervised and Self-supervised Speech Foundation Models2024FastAdaSP: Multitask-Adapted Efficient Inference for Large Speech Language Model2024