FunClip is an open-source video editing tool from Alibaba’s ModelScope community that transcribes video, generates subtitles, and helps you cut clips with LLM assistance. It has earned about 6,400 stars on GitHub and was actively updated in 2026.
FunClip is powered by FunASR, Alibaba’s speech recognition toolkit, and its Paraformer model, and it runs locally on your own machine through a Gradio interface — your footage does not need to leave your computer. The workflow is simple: transcribe a video, review the text, then select the segments worth keeping by speaker or by content and export them as a finished clip together with SRT subtitles. Speaker recognition based on CAM++ lets you cut by who is talking, hotword customization improves recognition of names and domain terms, and the LLM-assisted selection helps find highlights in long recordings. The 2.2.x series is the current line of development. The tool is especially popular with creators working in Chinese-language content, where Paraformer’s recognition quality is strong, but the pipeline is useful for any subtitling workflow. Everything — models, editing, and export — happens on the user’s own hardware, which also makes it a good fit for unpublished or sensitive footage.
FunClip is completely free and open source under the MIT license. There is no cloud tier; the tool runs locally with open models.
Video creators, educators, and podcasters who regularly turn long recordings into subtitled clips and want a private, offline workflow for doing it.