modelscope/FunClip

Project overview

FunClip automates video transcription, subtitle generation, and clipping locally using speech recognition and optional LLM-based content selection. It supports multi-segment clipping, speaker diarization, and hotword customization through a Gradio web UI or command line interface.

Project type

Video · Audio & Speech

Use cases

Translation & Subtitles · Video Creation

Deployment

Refer to project documentation

License

MIT

Best for

• Creators who need to process video files locally for transcription and multi-segment clipping.
• Developers or general users requiring automated text and LLM-based video highlight extraction.

Key capabilities

• Performs speech recognition on uploaded videos using Paraformer, Fun-ASR-Nano, or SenseVoice models to transcribe audio into text with timestamps.
• Extracts video highlights by combining ASR transcripts with LLMs like GPT or Qwen to automatically find and clip relevant segments.
• Integrates the CAM++ speaker recognition model to identify speakers, allowing users to clip segments from a specific speaker ID.
• Allows users to specify entity words or names as hotwords during the speech recognition process using SeACo-Paraformer to enhance recognition accuracy.
• Supports clipping multiple selected text segments freely and automatically returning the full video SRT subtitles and target segment SRT subtitles.
• Provides a Gradio-based web user interface that can be run locally or deployed on a server, accessed via a browser.
• Supports recognizing and clipping English audio files using the Paraformer English model or Whisper (planned).
• Provides a CLI tool to recognize and clip videos via commands.

Limitations and risks

• The Fun-ASR-Nano checkpoint does not provide reliable character-level timestamps; Paraformer is recommended for precise text-based clipping.
• Supporting the Whisper model for English requires massive GPU memory (planned).
• Clipping video files with embedded subtitles requires optional manual installation of ffmpeg and imagemagick.
• Using LLM-based clipping or the TwelveLabs API may incur costs or require paid API keys.

Getting started

• Clone the repository, install requirements via pip, and run the Gradio UI using python funclip/launch.py. The tool provides a simple pip installation and an automatic Gradio UI.

Evidence and sources

• GitHub project description: FunASR-powered video transcription, subtitle generation, and LLM-assisted clipping tool with a local Gradio UI.
• README: **FunClip** is a fully open-source, locally deployed automated video clipping tool.
• README: The functionalities are realized through Gradio interaction, offering simple installation and ease of use. It can also be deployed on a server and accessed via a browser.
• README: 🔥2024/05/13 FunClip v2.0.0 now supports smart clipping with large language models, integrating models from the qwen series, GPT series, etc., providing default prompts. You can al…
• README: FunClip now supports Fun-ASR-Nano and SenseVoice models. The `fun-asr-nano` option loads the flagship Fun-ASR-Nano-2512 checkpoint for Mandarin, English, Japanese, 7 Chinese diale…

Reviewed at: 2026-07-16T16:07:42Z