An LLM-driven multi-task system reports a 5.45% CER and 73.63% average SED F1 on the AS-70 Mandarin stuttering benchmark, though key baselines and uncertainty are missing.
<|im_start|>user<S><D><C ><P><|im_end|><|im_start|>assistant< T><|im_end|>|
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Leveraging LLM for Stuttering Speech: A Unified Architecture Bridging Recognition and Event Detection
An LLM-driven multi-task system reports a 5.45% CER and 73.63% average SED F1 on the AS-70 Mandarin stuttering benchmark, though key baselines and uncertainty are missing.