Add pipeline tag and update citation info

README.md

@@ -1,10 +1,12 @@
 ---
-license: mit
-datasets:
-- JacobLinCool/VoiceBank-DEMAND-16k
 base_model:
 - liduojia/MeanFlowSE
+datasets:
+- JacobLinCool/VoiceBank-DEMAND-16k
+license: mit
+pipeline_tag: audio-to-audio
 ---
+
 <div align="center">
 <p align="center">
   <h1>MeanFlowSE — One-Step Generative Speech Enhancement</h1>
@@ -88,7 +90,7 @@ Defaults assume 16 kHz audio, centered frames, Hann windows, and a complex STFT
 
 ```bash
 # Edit DATA_DIR and GPUs inside the script if needed
-bash train_vbd.sh
+bash scripts/train_vbd.sh
 ```
 
 Or run directly:
@@ -130,7 +132,7 @@ Use the helper script:
 MODE=onestep STEPS=1 \
 TEST_DATA_DIR=<BASE_DIR> \
 CKPT_INPUT=path/to/best.ckpt \
-bash run_inference.sh
+bash scripts/run_inference.sh
 ```
 
 Or call the evaluator:
@@ -221,10 +223,16 @@ We gratefully acknowledge **Prof. Xie Chen’s group (X-LANCE Lab, SJTU)** for t
 
 ## Citation
 
-* **Citation:** The paper is currently under review. We will add a BibTeX entry and article link once available.
-
-
-
-
-**Questions or issues?** Please open a GitHub issue or pull request.
-We welcome contributions — from bug fixes to new backbones and front-ends.
+Our article has been accepted by **ICASSP 2026**.
+
+```bibtex
+@misc{li2025meanflowseonestepgenerativespeech,
+  title         = {MeanFlowSE: one-step generative speech enhancement via conditional mean flow},
+  author        = {Duojia Li and Shenghui Lu and Hongchen Pan and Zongyi Zhan and Qingyang Hong and Lin Li},
+  year          = {2025},
+  eprint        = {2509.14858},
+  archivePrefix = {arXiv},
+  primaryClass  = {cs.SD},
+  url           = {https://arxiv.org/abs/2509.14858}
+}
+```