fix(sensevoice): stop rendering the Cough event as the Sneeze emoji

[MohammadHijjawi97]

Summary

rich_transcription_postprocess() maps SenseVoice event tags via event_dict, which had "<|Cough|>": "🤧" — the same emoji as <|Sneeze|> — so coughs were displayed as sneezes. emoji_dict and event_set in the same module already use 😷 for coughs (so format_str_v2 even strips 😷 spacing), which shows the intended mapping. This sets event_dict["<|Cough|>"] to 😷 in funasr/utils/postprocess_utils.py and in the identical copies shipped by funasr_onnx and funasr_torch.

Reproduction (before):

from funasr.utils.postprocess_utils import rich_transcription_postprocess
rich_transcription_postprocess("<|en|><|NEUTRAL|><|Cough|><|woitn|>hello")   # '🤧hello' (same as Sneeze)
# after: '😷hello'

Type of change

• Bug fix
• Documentation
• Example or demo
• Runtime or deployment
• Benchmark or evaluation
• Model/training change

Validation

• New tests/test_rich_transcription_postprocess.py (checks all three copies): 3 Cough cases fail on main, 6 passed with the fix.

• pytest tests/test_speaker_emotion_docs.py still passes.

• python -m compileall on the changed files; black (line-length 100) on the new test.

• python -m compileall funasr examples tests

• Docs or links checked

• Runtime/deployment command tested

User impact

SenseVoice users displaying rich transcriptions (demo scripts, ONNX/libtorch runtimes, apps built on rich_transcription_postprocess) can now tell coughs from sneezes.

Notes for reviewers

One-line mapping change per copy; no other tags affected.

[LauraGPT]

Verified exact head 78c9301580c977945649549f545b980e8ae2d370 against current main. The prior mapping renders Cough and Sneeze identically; the patch aligns all three shipped postprocessors with the existing cough emoji convention. Exact head and clean synthetic merge pass 6/6 new regressions and the adjacent speaker/emotion coverage. No blocking finding.