syncnet-python 0.2.1__tar.gz → 0.2.3__tar.gz

{syncnet_python-0.2.1 → syncnet_python-0.2.3}/INSTALL.md

@@ -11,8 +11,8 @@
 ### 1. Clone the repository
 
 ```bash
-git clone https://github.com/yourusername/SyncNet_py313.git
-cd SyncNet_py313
+git clone https://github.com/nawta/SyncNet_py309_313.git
+cd SyncNet_py309_313
 ```
 
 ### 2. Create virtual environment

{syncnet_python-0.2.1 → syncnet_python-0.2.3}/LICENSE

@@ -1,6 +1,7 @@
 MIT License
 
-Copyright (c) 2024 SyncNet Python 3.13 Contributors
+Copyright (c) 2016-present Joon Son Chung (original SyncNet code)
+Copyright (c) 2024 SyncNet Python 3.13 Contributors (modifications)
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

syncnet_python-0.2.3/LICENSE-APACHE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

{syncnet_python-0.2.1 → syncnet_python-0.2.3}/MANIFEST.in

@@ -1,19 +1,18 @@
 include README.md
 include LICENSE
+include LICENSE-APACHE
+include NOTICE
 include INSTALL.md
-include CLAUDE.md
 include requirements.txt
 
-# Include example files
-include example/*.wav
-include example/*.avi
+# Example media stay in the GitHub repository only; their source is not documented.
+prune example
 
 # Include scripts
 recursive-include scripts *.py
 
 # Include package data
 recursive-include syncnet *.py
-recursive-include script *.py
 
 # Exclude unnecessary files
 global-exclude __pycache__

syncnet_python-0.2.3/NOTICE

@@ -0,0 +1,43 @@
+syncnet-python
+https://github.com/nawta/SyncNet_py309_313
+
+This repository contains code under two licenses.
+
+1. MIT License (see LICENSE)
+   The original SyncNet code by Joon Son Chung
+   (https://github.com/joonson/syncnet_python) and the code written for
+   this repository: syncnet_python/__init__.py, syncnet_python/cli.py,
+   syncnet_python/safe_syncnet_utils.py,
+   syncnet_python/test_error_handling.py,
+   syncnet_python/test_lse_metrics.py, and the syncnet/, scripts/ and
+   config/ folders.
+
+2. Apache License 2.0 (see LICENSE-APACHE)
+   The following files come from MoChaBench
+   (https://github.com/congwei1230/MoChaBench, folder eval-lipsync/script/),
+   which is distributed under the Apache License 2.0. MoChaBench's LICENSE
+   file does not name a copyright holder; the files are credited here to
+   the MoChaBench authors. MoChaBench's versions are in turn based on the
+   original SyncNet code above. All files in this list except
+   syncnet_pipeline.py are unchanged copies of MoChaBench's files.
+
+     syncnet_python/syncnet_pipeline.py   (modified by nawta; the file
+                                           header lists the changes)
+     syncnet_python/SyncNetInstance.py
+     syncnet_python/SyncNetModel.py
+     syncnet_python/detectors/__init__.py
+     syncnet_python/detectors/s3fd/__init__.py
+     syncnet_python/detectors/s3fd/box_utils.py
+     syncnet_python/detectors/s3fd/nets.py
+     syncnet_python/run_syncnet_pipeline_on_1example.py
+     syncnet_python/run_syncnet_pipeline_on_mocha_generation_on_mocha_bench.py
+     syncnet_python/run_syncnet_pipeline_on_your_own_model_results.py
+
+Model weights
+
+   weights/syncnet_v2.model and weights/sfd_face.pth are copies of the files
+   published by the Visual Geometry Group, University of Oxford, at
+   https://www.robots.ox.ac.uk/~vgg/software/lipsync/. That page states:
+   "The model can be used for research purposes under Creative Commons
+   Attribution License." Neither the MIT License nor the Apache License 2.0
+   above applies to these files.

syncnet_python-0.2.3/PKG-INFO

@@ -0,0 +1,204 @@
+Metadata-Version: 2.4
+Name: syncnet-python
+Version: 0.2.3
+Summary: SyncNet: Audio-visual synchronization detection using deep learning. Updated version of https://github.com/joonson/syncnet_python for modern Python versions.
+Author: SyncNet Python Contributors
+Maintainer: SyncNet Python Contributors
+License-Expression: MIT AND Apache-2.0
+Project-URL: Homepage, https://github.com/nawta/SyncNet_py309_313
+Project-URL: Repository, https://github.com/nawta/SyncNet_py309_313
+Project-URL: Issues, https://github.com/nawta/SyncNet_py309_313/issues
+Keywords: audio-visual,synchronization,deep-learning,pytorch,lip-sync,video-processing,computer-vision
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Science/Research
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Multimedia :: Video
+Classifier: Topic :: Scientific/Engineering :: Image Processing
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+License-File: LICENSE-APACHE
+License-File: NOTICE
+Requires-Dist: torch>=2.0.0
+Requires-Dist: torchvision>=0.15.0
+Requires-Dist: numpy>=1.24.0
+Requires-Dist: scipy>=1.10.0
+Requires-Dist: pandas>=2.0.0
+Requires-Dist: scenedetect<0.7,>=0.6.0
+Requires-Dist: opencv-contrib-python>=4.8.0
+Requires-Dist: python-speech-features>=0.6
+Requires-Dist: ffmpeg-python>=0.2.0
+Provides-Extra: dev
+Requires-Dist: pytest>=7.4.0; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
+Requires-Dist: black>=23.0.0; extra == "dev"
+Requires-Dist: ruff>=0.1.0; extra == "dev"
+Requires-Dist: mypy>=1.7.0; extra == "dev"
+Requires-Dist: types-opencv-python; extra == "dev"
+Requires-Dist: types-scipy; extra == "dev"
+Dynamic: license-file
+
+# syncnet-python
+
+[![PyPI version](https://badge.fury.io/py/syncnet-python.svg)](https://badge.fury.io/py/syncnet-python)
+[![Python](https://img.shields.io/pypi/pyversions/syncnet-python.svg)](https://pypi.org/project/syncnet-python/)
+[![License: MIT AND Apache-2.0](https://img.shields.io/badge/License-MIT%20AND%20Apache--2.0-yellow.svg)](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE)
+
+A pip-installable SyncNet for Python 3.9 to 3.13 and PyTorch 2. It computes the confidence and minimum distance scores that talking-head and lip-sync papers report as LSE-C and LSE-D.
+
+SyncNet (Chung and Zisserman, 2016) is a network that measures how well mouth movements in a video match the speech audio. This package wraps the original SyncNet model and its S3FD face detector in one Python class and one command. Given a video, it finds and tracks faces, crops each face track, and returns three numbers per track: the audio-video offset in frames, the SyncNet confidence (LSE-C, higher means better sync), and the minimum audio-video feature distance (LSE-D, lower means better sync).
+
+## Quickstart
+
+### 1. Install
+
+```bash
+pip install syncnet-python
+```
+
+You also need `ffmpeg` on your `PATH` (`brew install ffmpeg` or `sudo apt-get install ffmpeg`).
+
+Use version 0.2.3 or later. Versions 0.2.2 and earlier accept scenedetect 0.7, which removed the `scenedetect.video_manager` module the pipeline imports, and then `from syncnet_python import SyncNetPipeline` silently gives `None`. Version 0.2.3 requires `scenedetect>=0.6,<0.7` and raises an `ImportError` with the underlying message if an import fails. If you must stay on 0.2.2, install it with `pip install syncnet-python==0.2.2 "scenedetect<0.7"`.
+
+The package uses `opencv-contrib-python` for `cv2`. Since 0.2.3 it is the only OpenCV package the install pulls in. Versions up to 0.2.2 also installed `opencv-python` through `scenedetect[opencv]`; if you upgrade an old environment and `cv2` fails to import, run `pip uninstall opencv-python opencv-contrib-python` and then `pip install opencv-contrib-python`.
+
+### 2. Download the weights
+
+The package does not include the model weights. Both files come from the Oxford VGG SyncNet page, the same URLs that the original repository's `download_model.sh` uses:
+
+```bash
+mkdir -p weights
+wget https://www.robots.ox.ac.uk/~vgg/software/lipsync/data/syncnet_v2.model -O weights/syncnet_v2.model
+wget https://www.robots.ox.ac.uk/~vgg/software/lipsync/data/sfd_face.pth -O weights/sfd_face.pth
+```
+
+The same two files are also in the [`weights/`](https://github.com/nawta/SyncNet_py309_313/tree/main/weights) folder of this repository. They are byte-identical to the Oxford files (SHA-256 `961e8696…5442` for `syncnet_v2.model` and `d54a87c2…c491` for `sfd_face.pth`).
+
+### 3. Python API
+
+```python
+from syncnet_python import SyncNetPipeline
+
+pipe = SyncNetPipeline(
+    {"s3fd_weights": "weights/sfd_face.pth", "syncnet_weights": "weights/syncnet_v2.model"},
+    device="cpu",  # or "cuda"
+)
+offsets, confs, dists, best_conf, min_dist, s3fd_json, has_face = pipe.inference(
+    video_path="clip.mp4",
+    audio_path=None,  # None uses the video's own audio track; or pass a .wav path
+)
+print("AV offset (frames):", offsets[0])
+print("LSE-C (confidence):", round(best_conf, 3))
+print("LSE-D (min distance):", round(min_dist, 3))
+```
+
+`offsets`, `confs` and `dists` are lists with one entry per face track. `best_conf` is the largest confidence over all tracks and `min_dist` is the smallest distance over all tracks. For a video with one face these equal `confs[0]` and `dists[0]`. With several faces the two values can come from different tracks, so use the per-track lists if you need scores for one speaker.
+
+`syncnet_python.calculate_lse_metrics(pipe, video_path)` returns `(lse_c, lse_d, quality_label)` from the same values.
+
+### 4. Command line
+
+```bash
+syncnet-python clip.mp4 --device cpu -o results.json
+```
+
+The command reads the weights from `weights/` in the current directory by default; change this with `--s3fd-weights` and `--syncnet-weights`. It prints the offset and confidence for each video and writes offset, confidence and minimum distance to the JSON file. The default device is `cuda`, so pass `--device cpu` on a machine without an NVIDIA GPU.
+
+### Tested setup
+
+The examples above were run on 2026-10-06 on an Apple Silicon Mac (CPU), using [`example/video.avi`](https://github.com/nawta/SyncNet_py309_313/blob/main/example/video.avi) converted to MP4. Each run used a fresh environment. The OpenCV column gives the version of `opencv-contrib-python`. In the 0.2.3 runs it was the only OpenCV package installed; in the 0.2.2 run `opencv-python` was also installed at the same version.
+
+| Package | Python | PyTorch | NumPy | OpenCV | scenedetect | Offset | LSE-C | LSE-D |
+|---|---|---|---|---|---|---|---|---|
+| 0.2.3 | 3.13 (arm64) | 2.14.1 | 2.5.3 | 5.0.0 | 0.6.7.1 | 1 | 4.529 | 9.237 |
+| 0.2.3 | 3.10 (arm64) | 2.14.1 | 2.2.6 | 5.0.0 | 0.6.7.1 | 1 | 4.529 | 9.237 |
+| 0.2.2 | 3.9 (x86_64 under Rosetta 2) | 2.2.2 | 1.26.4 | 4.11.0 | 0.6.7.1 | 1 | 4.524 | 9.291 |
+
+The 0.2.3 rows used the wheel built from this repository with a plain install. The Python API, `calculate_lse_metrics` and the CLI gave the same values. The Python 3.9 interpreter was an x86_64 build running under Rosetta 2. PyTorch 2.2.2 is the newest release with macOS x86_64 wheels, and it does not work with NumPy 2, so that run needed `numpy<2`, `opencv-python<4.12` and `opencv-contrib-python<4.12` installed by hand. The full run on the 5.3-second clip took about 10 seconds on CPU.
+
+## Changes in 0.2.3
+
+- Requires `scenedetect>=0.6,<0.7`, so a plain `pip install syncnet-python` works again.
+- A failed import inside the package raises `ImportError` with the original message. Earlier versions set `SyncNetPipeline` and the other exports to `None`.
+- The repository now holds the 0.2.2 code that was published on PyPI but not committed (`calculate_lse_metrics`, `audio_path=None`, `ffmpeg` error handling).
+- Package metadata declares the license as `MIT AND Apache-2.0` and ships `LICENSE`, `LICENSE-APACHE` and `NOTICE`.
+
+See [CHANGELOG.md](https://github.com/nawta/SyncNet_py309_313/blob/main/CHANGELOG.md) for earlier versions.
+
+## Comparison with joonson/syncnet_python
+
+The original repository, [joonson/syncnet_python](https://github.com/joonson/syncnet_python), was updated by its author on 2026-04-17 (PR #78). This table compares that version with this package.
+
+| | joonson/syncnet_python (2026-04) | syncnet-python 0.2.3 |
+|---|---|---|
+| Install | clone, then `conda env create -f environment.yml`; no `setup.py` or `pyproject.toml` | `pip install syncnet-python` |
+| Python | 3.10 (pinned in `environment.yml`) | 3.9 to 3.13 (3.9, 3.10 and 3.13 tested above) |
+| PyTorch | 2.5.1 (pinned) | `torch>=2.0.0` |
+| scenedetect | 0.6.7.1 (pinned) | `>=0.6,<0.7` |
+| Python API | `SyncNetInstance.evaluate()` and `extract_feature()` score a pre-cropped face clip; face detection, tracking and cropping run only through `run_pipeline.py` | `SyncNetPipeline(...).inference(video_path, audio_path)` runs face detection, tracking, cropping and scoring in one call |
+| Command line | `run_pipeline.py`, `run_syncnet.py`, `run_visualise.py` run in sequence, plus `demo_syncnet.py` for pre-cropped clips | one `syncnet-python` command that runs detection, tracking, cropping and scoring |
+| Output | offset, minimum distance and confidence written to the log; per-frame distances saved as `activesd.pckl` under `--data_dir` | values returned to Python, or written to JSON by the CLI |
+| Weights | downloaded by `download_model.sh` | downloaded separately (see above) |
+| Visualisation of the result | `run_visualise.py` | not included |
+
+## Errors this fixes
+
+Before the April 2026 update, the original repository pinned `scenedetect==0.5.1` and used `np.int`. Users hit these two errors in `run_pipeline.py`:
+
+```
+TypeError: 'tuple' object does not support item assignment
+```
+
+This comes from scenedetect 0.5.1's `ContentDetector` running with newer OpenCV (upstream issues [#55](https://github.com/joonson/syncnet_python/issues/55) and [#69](https://github.com/joonson/syncnet_python/issues/69)). This package uses scenedetect 0.6.
+
+```
+AttributeError: module 'numpy' has no attribute 'int'.
+```
+
+This comes from `.astype(np.int)` in `detectors/s3fd/box_utils.py`. NumPy 1.24 removed `np.int`. This package uses `.astype(int)`.
+
+## Repository layout
+
+The PyPI package contains only the `syncnet_python/` folder. Most of its files (`syncnet_pipeline.py`, `SyncNetInstance.py`, `SyncNetModel.py`, `detectors/` and the scripts named `run_syncnet_pipeline_on_*.py`) come from the SyncNet evaluation code in [MoChaBench](https://github.com/congwei1230/MoChaBench) (`eval-lipsync/script/`), which builds on the original repository. This package adds error handling around the `ffmpeg` calls in `syncnet_pipeline.py`, plus `cli.py`.
+
+The `syncnet/` folder holds a separate refactor with configuration files, logging and batch helpers. It is in this repository only and pip does not install it. `scripts/` has example scripts, and `example/` has a short test video with its audio.
+
+## Credits
+
+The SyncNet model, its pretrained weights and the original code are by Joon Son Chung and Andrew Zisserman ([joonson/syncnet_python](https://github.com/joonson/syncnet_python), [project page](https://www.robots.ox.ac.uk/~vgg/software/lipsync/)). The S3FD face detector weights (`sfd_face.pth`) are downloaded from the same page. The model, detector and pipeline files in `syncnet_python/` come from [MoChaBench](https://github.com/congwei1230/MoChaBench); [NOTICE](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE) lists them.
+
+## Citation
+
+If you use this code in your research, please cite the original paper:
+
+```bibtex
+@InProceedings{Chung16a,
+  author       = "Chung, J.~S. and Zisserman, A.",
+  title        = "Out of time: automated lip sync in the wild",
+  booktitle    = "Workshop on Multi-view Lip-reading, ACCV",
+  year         = "2016",
+}
+```
+
+## License
+
+This repository uses two licenses. [NOTICE](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE) lists which file falls under which.
+
+- The original SyncNet code by Joon Son Chung and the code written for this repository are under the MIT License ([LICENSE](https://github.com/nawta/SyncNet_py309_313/blob/main/LICENSE)).
+- The files taken from MoChaBench are under the Apache License 2.0 ([LICENSE-APACHE](https://github.com/nawta/SyncNet_py309_313/blob/main/LICENSE-APACHE)).
+
+The model weights have their own terms. The Oxford VGG SyncNet page says: "The model can be used for research purposes under Creative Commons Attribution License." The MIT and Apache licenses above do not cover the weights.
+
+## Links
+
+- Source and issues: https://github.com/nawta/SyncNet_py309_313
+- PyPI: https://pypi.org/project/syncnet-python/
+- Original SyncNet: https://github.com/joonson/syncnet_python

syncnet_python-0.2.3/README.md

@@ -0,0 +1,156 @@
+# syncnet-python
+
+[![PyPI version](https://badge.fury.io/py/syncnet-python.svg)](https://badge.fury.io/py/syncnet-python)
+[![Python](https://img.shields.io/pypi/pyversions/syncnet-python.svg)](https://pypi.org/project/syncnet-python/)
+[![License: MIT AND Apache-2.0](https://img.shields.io/badge/License-MIT%20AND%20Apache--2.0-yellow.svg)](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE)
+
+A pip-installable SyncNet for Python 3.9 to 3.13 and PyTorch 2. It computes the confidence and minimum distance scores that talking-head and lip-sync papers report as LSE-C and LSE-D.
+
+SyncNet (Chung and Zisserman, 2016) is a network that measures how well mouth movements in a video match the speech audio. This package wraps the original SyncNet model and its S3FD face detector in one Python class and one command. Given a video, it finds and tracks faces, crops each face track, and returns three numbers per track: the audio-video offset in frames, the SyncNet confidence (LSE-C, higher means better sync), and the minimum audio-video feature distance (LSE-D, lower means better sync).
+
+## Quickstart
+
+### 1. Install
+
+```bash
+pip install syncnet-python
+```
+
+You also need `ffmpeg` on your `PATH` (`brew install ffmpeg` or `sudo apt-get install ffmpeg`).
+
+Use version 0.2.3 or later. Versions 0.2.2 and earlier accept scenedetect 0.7, which removed the `scenedetect.video_manager` module the pipeline imports, and then `from syncnet_python import SyncNetPipeline` silently gives `None`. Version 0.2.3 requires `scenedetect>=0.6,<0.7` and raises an `ImportError` with the underlying message if an import fails. If you must stay on 0.2.2, install it with `pip install syncnet-python==0.2.2 "scenedetect<0.7"`.
+
+The package uses `opencv-contrib-python` for `cv2`. Since 0.2.3 it is the only OpenCV package the install pulls in. Versions up to 0.2.2 also installed `opencv-python` through `scenedetect[opencv]`; if you upgrade an old environment and `cv2` fails to import, run `pip uninstall opencv-python opencv-contrib-python` and then `pip install opencv-contrib-python`.
+
+### 2. Download the weights
+
+The package does not include the model weights. Both files come from the Oxford VGG SyncNet page, the same URLs that the original repository's `download_model.sh` uses:
+
+```bash
+mkdir -p weights
+wget https://www.robots.ox.ac.uk/~vgg/software/lipsync/data/syncnet_v2.model -O weights/syncnet_v2.model
+wget https://www.robots.ox.ac.uk/~vgg/software/lipsync/data/sfd_face.pth -O weights/sfd_face.pth
+```
+
+The same two files are also in the [`weights/`](https://github.com/nawta/SyncNet_py309_313/tree/main/weights) folder of this repository. They are byte-identical to the Oxford files (SHA-256 `961e8696…5442` for `syncnet_v2.model` and `d54a87c2…c491` for `sfd_face.pth`).
+
+### 3. Python API
+
+```python
+from syncnet_python import SyncNetPipeline
+
+pipe = SyncNetPipeline(
+    {"s3fd_weights": "weights/sfd_face.pth", "syncnet_weights": "weights/syncnet_v2.model"},
+    device="cpu",  # or "cuda"
+)
+offsets, confs, dists, best_conf, min_dist, s3fd_json, has_face = pipe.inference(
+    video_path="clip.mp4",
+    audio_path=None,  # None uses the video's own audio track; or pass a .wav path
+)
+print("AV offset (frames):", offsets[0])
+print("LSE-C (confidence):", round(best_conf, 3))
+print("LSE-D (min distance):", round(min_dist, 3))
+```
+
+`offsets`, `confs` and `dists` are lists with one entry per face track. `best_conf` is the largest confidence over all tracks and `min_dist` is the smallest distance over all tracks. For a video with one face these equal `confs[0]` and `dists[0]`. With several faces the two values can come from different tracks, so use the per-track lists if you need scores for one speaker.
+
+`syncnet_python.calculate_lse_metrics(pipe, video_path)` returns `(lse_c, lse_d, quality_label)` from the same values.
+
+### 4. Command line
+
+```bash
+syncnet-python clip.mp4 --device cpu -o results.json
+```
+
+The command reads the weights from `weights/` in the current directory by default; change this with `--s3fd-weights` and `--syncnet-weights`. It prints the offset and confidence for each video and writes offset, confidence and minimum distance to the JSON file. The default device is `cuda`, so pass `--device cpu` on a machine without an NVIDIA GPU.
+
+### Tested setup
+
+The examples above were run on 2026-10-06 on an Apple Silicon Mac (CPU), using [`example/video.avi`](https://github.com/nawta/SyncNet_py309_313/blob/main/example/video.avi) converted to MP4. Each run used a fresh environment. The OpenCV column gives the version of `opencv-contrib-python`. In the 0.2.3 runs it was the only OpenCV package installed; in the 0.2.2 run `opencv-python` was also installed at the same version.
+
+| Package | Python | PyTorch | NumPy | OpenCV | scenedetect | Offset | LSE-C | LSE-D |
+|---|---|---|---|---|---|---|---|---|
+| 0.2.3 | 3.13 (arm64) | 2.14.1 | 2.5.3 | 5.0.0 | 0.6.7.1 | 1 | 4.529 | 9.237 |
+| 0.2.3 | 3.10 (arm64) | 2.14.1 | 2.2.6 | 5.0.0 | 0.6.7.1 | 1 | 4.529 | 9.237 |
+| 0.2.2 | 3.9 (x86_64 under Rosetta 2) | 2.2.2 | 1.26.4 | 4.11.0 | 0.6.7.1 | 1 | 4.524 | 9.291 |
+
+The 0.2.3 rows used the wheel built from this repository with a plain install. The Python API, `calculate_lse_metrics` and the CLI gave the same values. The Python 3.9 interpreter was an x86_64 build running under Rosetta 2. PyTorch 2.2.2 is the newest release with macOS x86_64 wheels, and it does not work with NumPy 2, so that run needed `numpy<2`, `opencv-python<4.12` and `opencv-contrib-python<4.12` installed by hand. The full run on the 5.3-second clip took about 10 seconds on CPU.
+
+## Changes in 0.2.3
+
+- Requires `scenedetect>=0.6,<0.7`, so a plain `pip install syncnet-python` works again.
+- A failed import inside the package raises `ImportError` with the original message. Earlier versions set `SyncNetPipeline` and the other exports to `None`.
+- The repository now holds the 0.2.2 code that was published on PyPI but not committed (`calculate_lse_metrics`, `audio_path=None`, `ffmpeg` error handling).
+- Package metadata declares the license as `MIT AND Apache-2.0` and ships `LICENSE`, `LICENSE-APACHE` and `NOTICE`.
+
+See [CHANGELOG.md](https://github.com/nawta/SyncNet_py309_313/blob/main/CHANGELOG.md) for earlier versions.
+
+## Comparison with joonson/syncnet_python
+
+The original repository, [joonson/syncnet_python](https://github.com/joonson/syncnet_python), was updated by its author on 2026-04-17 (PR #78). This table compares that version with this package.
+
+| | joonson/syncnet_python (2026-04) | syncnet-python 0.2.3 |
+|---|---|---|
+| Install | clone, then `conda env create -f environment.yml`; no `setup.py` or `pyproject.toml` | `pip install syncnet-python` |
+| Python | 3.10 (pinned in `environment.yml`) | 3.9 to 3.13 (3.9, 3.10 and 3.13 tested above) |
+| PyTorch | 2.5.1 (pinned) | `torch>=2.0.0` |
+| scenedetect | 0.6.7.1 (pinned) | `>=0.6,<0.7` |
+| Python API | `SyncNetInstance.evaluate()` and `extract_feature()` score a pre-cropped face clip; face detection, tracking and cropping run only through `run_pipeline.py` | `SyncNetPipeline(...).inference(video_path, audio_path)` runs face detection, tracking, cropping and scoring in one call |
+| Command line | `run_pipeline.py`, `run_syncnet.py`, `run_visualise.py` run in sequence, plus `demo_syncnet.py` for pre-cropped clips | one `syncnet-python` command that runs detection, tracking, cropping and scoring |
+| Output | offset, minimum distance and confidence written to the log; per-frame distances saved as `activesd.pckl` under `--data_dir` | values returned to Python, or written to JSON by the CLI |
+| Weights | downloaded by `download_model.sh` | downloaded separately (see above) |
+| Visualisation of the result | `run_visualise.py` | not included |
+
+## Errors this fixes
+
+Before the April 2026 update, the original repository pinned `scenedetect==0.5.1` and used `np.int`. Users hit these two errors in `run_pipeline.py`:
+
+```
+TypeError: 'tuple' object does not support item assignment
+```
+
+This comes from scenedetect 0.5.1's `ContentDetector` running with newer OpenCV (upstream issues [#55](https://github.com/joonson/syncnet_python/issues/55) and [#69](https://github.com/joonson/syncnet_python/issues/69)). This package uses scenedetect 0.6.
+
+```
+AttributeError: module 'numpy' has no attribute 'int'.
+```
+
+This comes from `.astype(np.int)` in `detectors/s3fd/box_utils.py`. NumPy 1.24 removed `np.int`. This package uses `.astype(int)`.
+
+## Repository layout
+
+The PyPI package contains only the `syncnet_python/` folder. Most of its files (`syncnet_pipeline.py`, `SyncNetInstance.py`, `SyncNetModel.py`, `detectors/` and the scripts named `run_syncnet_pipeline_on_*.py`) come from the SyncNet evaluation code in [MoChaBench](https://github.com/congwei1230/MoChaBench) (`eval-lipsync/script/`), which builds on the original repository. This package adds error handling around the `ffmpeg` calls in `syncnet_pipeline.py`, plus `cli.py`.
+
+The `syncnet/` folder holds a separate refactor with configuration files, logging and batch helpers. It is in this repository only and pip does not install it. `scripts/` has example scripts, and `example/` has a short test video with its audio.
+
+## Credits
+
+The SyncNet model, its pretrained weights and the original code are by Joon Son Chung and Andrew Zisserman ([joonson/syncnet_python](https://github.com/joonson/syncnet_python), [project page](https://www.robots.ox.ac.uk/~vgg/software/lipsync/)). The S3FD face detector weights (`sfd_face.pth`) are downloaded from the same page. The model, detector and pipeline files in `syncnet_python/` come from [MoChaBench](https://github.com/congwei1230/MoChaBench); [NOTICE](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE) lists them.
+
+## Citation
+
+If you use this code in your research, please cite the original paper:
+
+```bibtex
+@InProceedings{Chung16a,
+  author       = "Chung, J.~S. and Zisserman, A.",
+  title        = "Out of time: automated lip sync in the wild",
+  booktitle    = "Workshop on Multi-view Lip-reading, ACCV",
+  year         = "2016",
+}
+```
+
+## License
+
+This repository uses two licenses. [NOTICE](https://github.com/nawta/SyncNet_py309_313/blob/main/NOTICE) lists which file falls under which.
+
+- The original SyncNet code by Joon Son Chung and the code written for this repository are under the MIT License ([LICENSE](https://github.com/nawta/SyncNet_py309_313/blob/main/LICENSE)).
+- The files taken from MoChaBench are under the Apache License 2.0 ([LICENSE-APACHE](https://github.com/nawta/SyncNet_py309_313/blob/main/LICENSE-APACHE)).
+
+The model weights have their own terms. The Oxford VGG SyncNet page says: "The model can be used for research purposes under Creative Commons Attribution License." The MIT and Apache licenses above do not cover the weights.
+
+## Links
+
+- Source and issues: https://github.com/nawta/SyncNet_py309_313
+- PyPI: https://pypi.org/project/syncnet-python/
+- Original SyncNet: https://github.com/joonson/syncnet_python