AIGCPanel is an open-source desktop application for producing AI digital human content. Digital human video synthesis, speech synthesis and voice cloning, speech recognition, an audio-video toolbox and model management all live in one client, and the models can be downloaded to run locally rather than rented from a cloud. The source is published on GitHub and Gitee under the Apache-2.0 licence.
Its centrepiece is lip sync. You supply a person video or a photo plus either text or an audio file, and a lip-sync model produces a talking clip; text is first voiced by a TTS model, which then drives the picture. The site names MuseTalk, LatentSync, Wav2Lip and Heygem, while the model market also lists SadTalker, LivePortrait and SoulX-FlashHead, each with its own VRAM and platform requirements. Green-screen footage is supported and avatar templates can be stored and reused.
On the audio side, several TTS engines are available — CosyVoice, FishSpeech, IndexTTS, GPT-SoVITS, Spark-TTS and others — with adjustable speed and pitch, zero-shot voice cloning from a reference clip, and batch downloads from the task list. Speech recognition returns timestamped text that can be exported as subtitles, alongside voice replacement and audio normalisation. The toolbox adds more than 25 utilities covering long-text to audio, subtitle to audio, video backgrounds, smart trimming, subtitles, speed, compression, format conversion, intro and outro cards, merging, custom FFmpeg commands, text-to-image, image-to-image and background removal.
The entry bar is modest. The interface itself needs no GPU, while running models locally is best with an NVIDIA RTX 3060 or better and 8GB of VRAM; without a discrete card you can use cloud models billed by consumption through prepaid credits. Windows 10+, macOS 10.15+ on Intel or Apple silicon, and Linux distributions such as UOS, Ubuntu and Deepin are supported, with a Simplified Chinese or English interface and a dark mode. The community edition is free and open source; live streaming with danmaku interaction across Douyin, Bilibili, Huya, Douyu and Kuaishou, visual workflows and cloud models belong to the Pro/VIP tiers, whose prices are not published on the site and must be confirmed in the app or with support. Compliance matters too: the site forbids unlawful use, its disclaimer states that bundled resources are for personal study, commercial use requires you to clear model and footage rights yourself, and cloning someone's voice or likeness needs their consent.

A1: It generates lip-synced digital human videos, synthesises and clones voices, transcribes speech into timestamped subtitles, and adds 25+ audio-video tools, visual workflows and multi-platform danmaku live interaction.
A2: The community edition is free and open source under Apache-2.0 and covers the core creation features, while Pro/VIP adds live streaming, visual workflows and cloud models; prices are not published on the site and should be confirmed in the app or with support.
A3: Yes. The interface does not need a GPU, and cloud models are billed by usage; for local inference the site recommends an NVIDIA RTX 3060 or better with at least 8GB of VRAM.
A4: Windows 10+, macOS 10.15+ on Intel or Apple silicon, and Linux distributions such as UOS, Ubuntu and Deepin, with 64-bit and ARM64 builds, plus a Simplified Chinese or English interface.
A5: The application is Apache-2.0 licensed, but the site's disclaimer states that bundled resources are for personal study, so commercial use requires you to clear rights in the models, footage, likeness and voice and to follow local law.
A6: Download a one-click runtime pack from the model market and start it, or follow the documentation to connect a third-party model yourself, or use remote API and cloud model services.