Speech AI vs human speaker similarity study

Arxiv CS.SD (Sound/Audio ML)

Study compares speaker embeddings from speech foundation models to human perception of speaker similarity. Listeners judged similarity on a continuous scale, evaluated against model embeddings.

speechvoice-ai

Arxiv CS.SD (Sound/Audio ML)·Minoru KishiHayato YagiShinnosuke TakamichiYuki Saito

6 days ago