Commit History
3 days ago
-
Replace the training-data pipeline figure
Verified signatureRender the new pipeline figure (speech generation with Qwen3-TTS, an ASR check and Qwen3-ForcedAligner; online training-data simulation) from its PDF at 2400 px wide, and describe it in the alt text. Co-Authored-By: Claude Code <noreply@anthropic.com>d962636 -
Add the training-data pipeline figure
Verified signatureAdd a Training data section with the paper's data-generation pipeline figure (assets/data-pipeline.png, a copy of pipeline.png drawn by make_figures.py) and a short description of the source speech, Qwen3-TTS synthesis, spk-mix composition, shortcut control and audit, and online augmentation. Co-Authored-By: Claude Code <noreply@anthropic.com>69a175e -
Add the midea-test results table and bar chart
Verified signatureMain results from the paper: the midea-test table and the bar chart of pos-CER, RR and the unified score. Also state the TS-Aishell cue as the first 2-5 characters and drop heavy noise from the untested conditions.5d93a70 -
Add a PNG rendering of the architecture figure (cards cannot embed PDFs)
Verified signatureCo-Authored-By: Claude Code <noreply@anthropic.com>6008d54 -
Upload model.pdf
Verified signature689c033 -
Delete assets/architecture.png
Verified signature2eae262 -
Drop the old overview figure, replaced by wakecue-asr.png
Verified signatureCo-Authored-By: Claude Code <noreply@anthropic.com>8382ca0 -
Add assets/architecture.png
Verified signatureCo-Authored-By: Claude Code <noreply@anthropic.com>6015551 -
Add assets/wakecue-asr.png
Verified signatureCo-Authored-By: Claude Code <noreply@anthropic.com>331f848
13 days ago
-
Add overview figure to the model card
Verified signature1eed9de