analiza aiqbench
This study demonstrates that injecting speaker embeddings via Feature-wise Linear Modulation (FiLM) into a frozen ASR encoder effectively adapts models to pathological speech without altering base weights. The approach achieves performance competitive with established fine-tuning strategies while maintaining robustness on standard speech inputs. For technical audiences, this offers a parameter-efficient pathway to improve accessibility in speech recognition systems for neurological conditions.
FiLM-Based Speaker Conditioning of a SpeechLLM for Pathological Speech Recognition
arXiv:2606.06211v1 Announce Type: new Abstract: Automatic speech recognition (ASR) has advanced remarkably for standard speech; however, pathological speech from neurological conditions remains a significant challenge. We investigate speaker conditioning via Feature-wise Linear Modulation (FiLM), injecting x-vector-derived information into each transformer layer of a frozen ASR encoder to adapt internal representations to individual pathological speakers without modifying base model weights. We benchmark this for the ASR task against standard and parameter-efficient fine-tuning baselines, complemented by post-processing, on Spanish and English pathological speech. Additionally, we evaluate if the adapted model preserves the ability to answer speech-related questions. Results show that speaker-conditioned ASR is competitive with established adaptation strategies while retaining performance on non-conditioned speech.
Abstract:Automatic speech recognition (ASR) has advanced remarkably for standard speech; however, pathological speech from neurological conditions remains a significant challenge. We investigate speaker conditioning via Feature-wise Linear Modulation (FiLM), injecting x-vector-derived information into each transformer layer of a frozen ASR encoder to adapt internal representations to individual pathological speakers without modifying base model weights. We benchmark this for the ASR task against standard and parameter-efficient fine-tuning baselines, complemented by post-processing, on Spanish and English pathological speech. Additionally, we evaluate if the adapted model preserves the ability to answer speech-related questions. Results show that speaker-conditioned ASR is competitive with established adaptation strategies while retaining performance on non-conditioned speech.
Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS)
arXiv:2606.06211 [cs.CL]
(or arXiv:2606.06211v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.06211 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 209 cuvinte
Subiecte conexe
Controllable Dysarthric Speech Synthesis with Patient-Specific Conditioning for Speaker-Diverse ASR Augmentation
arXiv · 15 sept. 2026
Asymmetric Classifier-Free Guidance for Target-Speaker ASR
arXiv · 28 sept. 2026
A New Transformer-Based Approach for Audio-Based Kinship Verification and a New Uncontrolled Mandarin Kinship Speech Dataset
arXiv · 16 sept. 2026
**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**
Hugging Face · 23 sept. 2026