建议添加 FunASR/SenseVoice 作为语音识别引擎

[!NOTE] License and capability correction (2026-07-14): FunASR is a toolkit, not a single model. The source code of FunASR and SenseVoice repositories is licensed under MIT; the model weights are based on the model cards. SenseVoiceSmall supports Chinese, Cantonese, English, Japanese, and Korean, and the weights use the open source agreement of the FunASR model linked to the model card; Fun-ASR-Nano-2512 is licensed under Apache-2.0. The language range, punctuation, and performance depend on the selected model and the runtime configuration.