FiLM-Based Speaker Conditioning of a SpeechLLM for Pathological Speech Recognition

Arxiv CS.SD (Sound/Audio ML)

The paper applies Feature-wise Linear Modulation (FiLM) to condition a SpeechLLM for pathological speech recognition, targeting the challenge of ASR for neurological conditions. The method aims to improve performance on non-standard speech patterns.

llmspeech-to-text

Arxiv CS.SD (Sound/Audio ML)·Fernando L\'opezSantosh KesirajuJordi Luque

6 days ago