BILIVE - Open-source AI recording tool, one-stop automatic live streaming recording and processing

What is BILIVE

BILIVE is an open-source tool based on AI technology, designed specifically for Bilibili live streaming recording and processing. The tool supports automatic live streaming recording, rendering danmaku and subtitles, supports speech recognition, automatic clipping of highlight segments, and generates interesting titles and stylized video covers. BILIVE can automatically submit processed videos to Bilibili, integrates multiple modality models, is compatible with ultra-low-spec machines, and can run without a GPU, making it suitable for individual users and small servers.

Main features of BILIVE

• Live recording: Automatically record Bilibili live streams, support simultaneous recording of multiple rooms, including danmaku and gift information.
• Danmaku and subtitles: Automatically render danmaku, support speech recognition to generate subtitles.
• Automatic clipping: Extract highlight segments based on danmaku density and generate titles.
• Cover generation: Automatically generate stylized video covers.
• Automatic submission: Support submitting videos to Bilibili, compatible with multi-part uploads.
• Low hardware requirements: No GPU required, suitable for running on ultra-low-spec machines.

How to use BILIVE

• Clone the project: Run the following commands in the terminal to clone the project and initialize submodules:

git clone --recurse-submodules https://github.com/timerring/bilive.git
cd bilive

• Install dependencies: Install Python dependencies in the project directory, and make sure ffmpeg is installed:

pip install -r requirements.txt

• Configuration and startup：
  • Core configuration: Set the API Key and mode for speech recognition (ASR), clipping model (MLLM), and cover generation (Image Gen). Configure video upload information (title, description template, partition tid, etc.).
  • Recording configuration: Add the room_id of the room to be recorded. Set parameters such as recording duration duration_limit.
  • Login：
    • Log in using bilitool：

bilitool login --export

• • • Or log in within the submodule：

cd src/upload/bilitool
python -m bilitool.cli login

• • Set the recording page password: Set the environment variable RECORD_KEY.

export RECORD_KEY=your_password

• • Start recording：

./record.sh

• • Start processing and uploading：

./upload.sh

BILIVE project address

• GitHub repository：https://github.com/timerring/bilive

BILIVE use cases

• UP creator content creation: Automatically record live streams, generate danmaku, subtitles, clips, and covers, and quickly submit to Bilibili.
• Multi-platform streaming: Support 24/7 looping streams and manage multiple live rooms simultaneously.
• Highlight segment extraction: Intelligently identify high-energy segments, generate attractive titles, and help with content promotion.
• Low-spec friendly: No GPU required, suitable for running on old devices, with automatic cleanup to save space.
• Learning and research: Recorded data can be used for analysis, and the open-source code is convenient for technical learning.