Audio/video transcription and summarization tool based on faster-whisper, suitable for Windows environment.
If you don't like this background image, you can replace it and adjust the transparency as you wish.
Program Entry
python -m src.ui.gui1. Download and Dependencies
Download the portable version from the release page, extract it, and double-click to run. The program will automatically download dependencies.
If the download fails, check if the download source exists:
- faster-whisper-large-v3-turbo-ct2 (transcription speech model)
- cuBLAS.and.cuDNN_CUDA12_win_v3.7z (GPU acceleration dependency libraries)
libs directory description:
Normally automatically downloads the cuBLAS.and.cuDNN_CUDA12_win_v3.7z package and automatically extracts it to the libs directory, containing CUDA 12 and cuDNN 9 dynamic libraries used for local GPU-accelerated speech recognition inference:
video2text\libs\
├── cublas64_12.dll
├── cublasLt64_12.dll
├── cudnn64_9.dll
├── cudnn_adv64_9.dll
├── cudnn_cnn64_9.dll
├── cudnn_engines_precompiled64_9.dll
├── cudnn_engines_runtime_compiled64_9.dll
├── cudnn_graph64_9.dll
├── cudnn_heuristic64_9.dll
├── cudnn_ops64_9.dll
├── readme_en.md
└── readme_zh.md
models directory structure:
Normally automatically downloads the faster-whisper-large-v3-turbo-ct2 model. If you want to use other models, you can download them from Systran
video2text\models\
└── faster-whisper-large-v3-turbo-ct2\
├── config.json
├── model.bin
├── preprocessor_config.json
├── README.md
├── tokenizer.json
├── vocabulary.json
└── .gitattributes
After placing the models, you can use the audio/video to text feature.
2. Summary Model Installation
Supports two summary services: NVIDIA online models and local Ollama models. Choose one as needed. Configuration steps: Settings > Edit Config > Summary > Radio button selection
2.1 NVIDIA Online (Use online NVIDIA models for summarization)
You need to first register on NVIDIA Build and create an API Key, then add it to the program via Settings > Manage API Key
2.2 Install Ollama (Use local models for summarization)
Ollama is a local large language model runtime framework used for generating text summaries. If you have limited VRAM, it may be difficult to find a model with good summarization performance. It is recommended to use online models directly.
- For download and installation, see the official website. You need to add ollama to the environment variables
- The ollama service will be automatically started during summarization, and there are start and stop functions in [Configuration]
After registering for an Ollama account, there are also free online models you can use
2.3 mistral Online Model
You need to first register on mistral, then create an API Key, and add it to the program via Settings > Manage API Key