Nguồn & giấy phép
🔒 Video không rời khỏi máy bạn. Mọi xử lý chạy ngay trong trình duyệt; trang chỉ tải mã của ứng dụng và các file mô hình.
Nơi lưu mô hình
Bản công khai tải các file mô hình (.onnx) từ một kho mô hình công khai trên Hugging Face: huggingface.co/minhducjp/webdonhac-models. Trình duyệt kiểm tra mã sha256 của từng file với models.json rồi lưu trong bộ nhớ đệm (Cache Storage) của trình duyệt; video của bạn không bao giờ được gửi lên. Bản chạy trên máy (serve.py) lấy cùng các file này từ thư mục --models trên máy bạn.
| File | Chế độ | Dung lượng | Nguồn gốc | Giấy phép |
|---|---|---|---|---|
mn10_as.onnx | Nhanh | 24.0 MB | EfficientAT (Florian Schmid, Khaled Koutini, Gerhard Widmer) | MIT |
panns_mnv2.onnx | Nhanh | 20.7 MB | PANNs (Qiuqiang Kong và cộng sự) | CC BY 4.0 |
mn40_as_ext.onnx | Chính xác | 278.1 MB | EfficientAT (Florian Schmid, Khaled Koutini, Gerhard Widmer) | MIT |
panns_cnn14.onnx | Chính xác | 327.3 MB | PANNs (Qiuqiang Kong và cộng sự) | CC BY 4.0 |
Cả bốn mô hình được huấn luyện trên AudioSet; tên 527 nhãn đầu ra theo AudioSet ontology (CC BY-SA 4.0, xem bên dưới).
PANNs: mô hình CNN14 và MobileNetV2
Qiuqiang Kong, Yin Cao, Turab Iqbal, Yuxuan Wang, Wenwu Wang, Mark D. Plumbley. “PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition.” IEEE/ACM Transactions on Audio, Speech, and Language Processing, 28 (2020), 2880–2894. arXiv:1912.10211 · mã nguồn · trọng số đã huấn luyện (Zenodo).
Trọng số dùng theo giấy phép Creative Commons Attribution 4.0 (CC BY 4.0). Thay đổi: Cnn14_mAP=0.431 và MobileNetV2_mAP=0.383 đã được chuyển sang ONNX (kèm phần trích đặc trưng) để chạy trong trình duyệt; không huấn luyện lại.
EfficientAT: mô hình mn10_as và mn40_as_ext
Florian Schmid, Khaled Koutini, Gerhard Widmer. “Efficient Large-Scale Audio Tagging via Transformer-to-CNN Knowledge Distillation.” ICASSP 2023. github.com/fschmid56/EfficientAT. Đã chuyển sang ONNX. Giấy phép của kho EfficientAT:
MIT License Copyright (c) 2022 Florian Schmid Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Nhãn âm thanh
Danh sách 527 nhãn AudioSet (labels.csv) lấy từ AudioSet ontology của Google, giấy phép Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0).
Thư viện chạy trong trình duyệt
- ONNX Runtime Web 1.30.0, Microsoft. MIT License (toàn văn).
- Mediabunny 1.61.0, Vanilagy. Mozilla Public License 2.0 (toàn văn). Dùng nguyên bản, không sửa.