WavLM
Self-supervised speech model pre-trained on 94k hours for full-stack speech tasks.
About
WavLM is a large-scale self-supervised speech model from Microsoft, released in October 2021. Pre-trained on 94,000 hours of public audio, it extends the HuBERT framework with masked speech prediction and denoising modeling so one backbone serves both speech recognition and non-ASR tasks.
WavLM set state-of-the-art results on the SUPERB benchmark and on representative test sets for speaker verification, speech separation, and speaker diarization. It became a standard backbone for speaker-related research because of its strong identity modeling.
Microsoft released the code and Base, Base+, and Large checkpoints, which are widely used for fine-tuning speech applications.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 86; Trending score 60/100 from researcher assessment.