FunClip
Automated video clipping and subtitle generation using FunASR and LLMs.
Transcribe, diarize, and clip videos using local AI speech models and LLMs.
FunClip is an open-source tool that automates video clipping and subtitle generation through a local Gradio UI. It uses Alibaba's FunASR models and LLMs to let you cut videos by selecting text segments, speakers, or hotwords.
What it is
The software performs speech recognition via Paraformer or SenseVoice models to generate precise timestamps and SRT subtitles. You can then click through a browser interface to extract clips based on specific words, phrases, or recognized speakers.
Who it's for
It suits video editors and content creators looking to automate transcription-based clipping locally. It does not suit users wanting a traditional timeline-based NLE without command-line setup.
Key features
FunASR Integration
Leverages Paraformer and SenseVoice models for high-accuracy speech recognition and timestamp prediction.
Speaker Diarization
Uses CAM++ or MOSS backends to identify speaker IDs and clip segments belonging to a specific voice.
Hotword Customization
Allows specifying custom entity names and vocabulary to improve recognition accuracy during transcription.
Local Gradio UI
Provides a browser-based interface for easy deployment, interaction, and multi-segment video trimming.
Up and running in 3 steps
Our honest verdict
What's great
- Accurate timestamp prediction integrated into Chinese and multilingual ASR models.
- Speaker-based clipping using built-in identification models.
- Hotword customization for specific entity and name recognition.
- Automatic generation of full and target segment SRT subtitles.
Where it falls short
- Requires managing a Python virtual environment and specific PyTorch dependencies.
- Certain model backends like MOSS require setting up external vLLM services.
- High memory requirements for advanced models and timestamp generation.
Pricing & license
Free under its MIT licence. There are no paid hosted options mentioned in the README, so the only cost is running it on your own hardware.
Open remote AI-training roles
874 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
What models does FunClip support for ASR?
It supports Paraformer-Large, Fun-ASR-Nano, SenseVoice, and MOSS for long-form transcription.
Can I clip videos based on a specific speaker?
Yes, it integrates speaker recognition models to let you target trimming by speaker ID.
Do I need ImageMagick installed?
ImageMagick is optional unless you are running legacy subtitle tests or custom MoviePy workflows.
