Skip to content

GitLab

Commit 5c023842 authored by chenpangpang's avatar chenpangpang
Browse files

feat: 增加LatentSync

parent 822b66ca
Pipeline #2211 canceled with stages
Hide whitespace changes
Inline Side-by-side
Showing with 978 additions and 1 deletion
LatentSync @ 9c9dbc30
Subproject commit 9c9dbc30b3cefb17111907c43ade65830f2883d8
LatentSync/LICENSE 0 → 100644
View file @ 5c023842
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.
\ No newline at end of file
LatentSync/ORIGINAL_README.md 0 → 100644
View file @ 5c023842
# LatentSync: Audio Conditioned Latent Diffusion Models for Lip Sync
<div align="center">
[![arXiv](https://img.shields.io/badge/arXiv_paper-2412.09262-b31b1b)](https://arxiv.org/abs/2412.09262)
</div>
## 📖 Abstract
We present *LatentSync*, an end-to-end lip sync framework based on audio conditioned latent diffusion models without any intermediate motion representation, diverging from previous diffusion-based lip sync methods based on pixel space diffusion or two-stage generation. Our framework can leverage the powerful capabilities of Stable Diffusion to directly model complex audio-visual correlations. Additionally, we found that the diffusion-based lip sync methods exhibit inferior temporal consistency due to the inconsistency in the diffusion process across different frames. We propose *Temporal REPresentation Alignment (TREPA)* to enhance temporal consistency while preserving lip-sync accuracy. TREPA uses temporal representations extracted by large-scale self-supervised video models to align the generated frames with the ground truth frames.
## 🏗️ Framework
<p align="center">
<img src="assets/framework.png" width=100%>
<p>
LatentSync uses the Whisper to convert melspectrogram into audio embeddings, which are then integrated into the U-Net via cross-attention layers. The reference and masked frames are channel-wise concatenated with noised latents as the input of U-Net. In the training process, we use one-step method to get estimated clean latents from predicted noises, which are then decoded to obtain the estimated clean frames. The TREPA, LPIPS and SyncNet loss are added in the pixel space.
## 🎬 Demo
<table class="center">
<tr style="font-weight: bolder;text-align:center;">
<td width="50%"><b>Original video</b></td>
<td width="50%"><b>Lip-synced video</b></td>
</tr>
<tr>
<td>
<video src=https://github.com/user-attachments/assets/ff3a84da-dc9b-498a-950f-5c54f58dd5c5 controls preload></video>
</td>
<td>
<video src=https://github.com/user-attachments/assets/150e00fd-381e-4421-a478-a9ea3d1212a8 controls preload></video>
</td>
</tr>
<tr>
<td>
<video src=https://github.com/user-attachments/assets/32c830a9-4d7d-4044-9b33-b184d8e11010 controls preload></video>
</td>
<td>
<video src=https://github.com/user-attachments/assets/84e4fe9d-b108-44a4-8712-13a012348145 controls preload></video>
</td>
</tr>
<tr>
<td>
<video src=https://github.com/user-attachments/assets/7510a448-255a-44ee-b093-a1b98bd3961d controls preload></video>
</td>
<td>
<video src=https://github.com/user-attachments/assets/6150c453-c559-4ae0-bb00-c565f135ff41 controls preload></video>
</td>
</tr>
<tr>
<td width=300px>
<video src=https://github.com/user-attachments/assets/0f7f9845-68b2-4165-bd08-c7bbe01a0e52 controls preload></video>
</td>
<td width=300px>
<video src=https://github.com/user-attachments/assets/c34fe89d-0c09-4de3-8601-3d01229a69e3 controls preload></video>
</td>
</tr>
<tr>
<td>
<video src=https://github.com/user-attachments/assets/7ce04d50-d39f-4154-932a-ec3a590a8f64 controls preload></video>
</td>
<td>
<video src=https://github.com/user-attachments/assets/70bde520-42fa-4a0e-b66c-d3040ae5e065 controls preload></video>
</td>
</tr>
</table>
(Photorealistic videos are filmed by contracted models, and anime videos are from [VASA-1](https://www.microsoft.com/en-us/research/project/vasa-1/) and [EMO](https://humanaigc.github.io/emote-portrait-alive/))
## 📑 Open-source Plan
- [x] Inference code and checkpoints
- [x] Data processing pipeline
- [x] Training code
## 🔧 Setting up the Environment
Install the required packages and download the checkpoints via:
```bash
source setup_env.sh
```
If the download is successful, the checkpoints should appear as follows:
```
./checkpoints/
|-- latentsync_unet.pt
|-- latentsync_syncnet.pt
|-- whisper
| `-- tiny.pt
|-- auxiliary
| |-- 2DFAN4-cd938726ad.zip
| |-- i3d_torchscript.pt
| |-- koniq_pretrained.pkl
| |-- s3fd-619a316812.pth
| |-- sfd_face.pth
| |-- syncnet_v2.model
| |-- vgg16-397923af.pth
| `-- vit_g_hybrid_pt_1200e_ssv2_ft.pth
```
These already include all the checkpoints required for latentsync training and inference. If you just want to try inference, you only need to download `latentsync_unet.pt` and `tiny.pt` from our [HuggingFace repo](https://huggingface.co/chunyu-li/LatentSync)
## 🚀 Inference
Run the script for inference, which requires about 6.5 GB GPU memory.
```bash
./inference.sh
```
You can change the parameter `guidance_scale` to 1.5 to improve the lip-sync accuracy.
## 🔄 Data Processing Pipeline
The complete data processing pipeline includes the following steps:
1. Remove the broken video files.
2. Resample the video FPS to 25, and resample the audio to 16000 Hz.
3. Scene detect via [PySceneDetect](https://github.com/Breakthrough/PySceneDetect).
4. Split each video into 5-10 second segments.
5. Remove videos where the face is smaller than 256 $\times$ 256, as well as videos with more than one face.
6. Affine transform the faces according to the landmarks detected by [face-alignment](https://github.com/1adrianb/face-alignment), then resize to 256 $\times$ 256.
7. Remove videos with [sync confidence score](https://www.robots.ox.ac.uk/~vgg/publications/2016/Chung16a/chung16a.pdf) lower than 3, and adjust the audio-visual offset to 0.
8. Calculate [hyperIQA](https://openaccess.thecvf.com/content_CVPR_2020/papers/Su_Blindly_Assess_Image_Quality_in_the_Wild_Guided_by_a_CVPR_2020_paper.pdf) score, and remove videos with scores lower than 40.
Run the script to execute the data processing pipeline:
```bash
./data_processing_pipeline.sh
```
You can change the parameter `input_dir` in the script to specify the data directory to be processed. The processed data will be saved in the same directory. Each step will generate a new directory to prevent the need to redo the entire pipeline in case the process is interrupted by an unexpected error.
## 🏋️‍♂️ Training U-Net
Before training, you must process the data as described above and download all the checkpoints. We released a pretrained SyncNet with 94% accuracy on the VoxCeleb2 dataset for the supervision of U-Net training. Note that this SyncNet is trained on affine transformed videos, so when using or evaluating this SyncNet, you need to perform affine transformation on the video first (the code of affine transformation is included in the data processing pipeline).
If all the preparations are complete, you can train the U-Net with the following script:
```bash
./train_unet.sh
```
You should change the parameters in U-Net config file to specify the data directory, checkpoint save path, and other training hyperparameters.
## 🏋️‍♂️ Training SyncNet
In case you want to train SyncNet on your own datasets, you can run the following script. The data processing pipeline for SyncNet is the same as U-Net.
```bash
./train_syncnet.sh
```
After `validations_steps` training, the loss charts will be saved in `train_output_dir`. They contain both the training and validation loss.
LatentSync/README.md 0 → 100644
View file @ 5c023842
---
title: LatentSync
emoji: 👄
colorFrom: blue
colorTo: blue
sdk: gradio
sdk_version: 5.9.1
app_file: app.py
pinned: false
short_description: Audio Conditioned LipSync with Latent Diffusion Models
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
\ No newline at end of file
LatentSync/app.py 0 → 100644
View file @ 5c023842
import gradio as gr
import os
import sys
import shutil
import uuid
import subprocess
from glob import glob
from huggingface_hub import snapshot_download
# Download models
os.makedirs("checkpoints", exist_ok=True)
snapshot_download(
repo_id = "chunyu-li/LatentSync",
local_dir = "./checkpoints"
)
import tempfile
from moviepy.editor import VideoFileClip
from pydub import AudioSegment
def process_video(input_video_path, temp_dir="temp_dir"):
"""
Crop a given MP4 video to a maximum duration of 10 seconds if it is longer than 10 seconds.
Save the new video in the specified folder (default is temp_dir).
Args:
input_video_path (str): Path to the input video file.
temp_dir (str): Directory where the processed video will be saved.
Returns:
str: Path to the cropped video file.
"""
# Ensure the temp_dir exists
os.makedirs(temp_dir, exist_ok=True)
# Load the video
video = VideoFileClip(input_video_path)
# Determine the output path
input_file_name = os.path.basename(input_video_path)
output_video_path = os.path.join(temp_dir, f"cropped_{input_file_name}")
# Crop the video to 10 seconds if necessary
if video.duration > 10:
video = video.subclip(0, 10)
# Write the cropped video to the output path
video.write_videofile(output_video_path, codec="libx264", audio_codec="aac")
# Return the path to the cropped video
return output_video_path
def process_audio(file_path, temp_dir):
# Load the audio file
audio = AudioSegment.from_file(file_path)
# Check and cut the audio if longer than 4 seconds
max_duration = 8 * 1000 # 4 seconds in milliseconds
if len(audio) > max_duration:
audio = audio[:max_duration]
# Save the processed audio in the temporary directory
output_path = os.path.join(temp_dir, "trimmed_audio.wav")
audio.export(output_path, format="wav")
# Return the path to the trimmed file
print(f"Processed audio saved at: {output_path}")
return output_path
import argparse
from omegaconf import OmegaConf
import torch
from diffusers import AutoencoderKL, DDIMScheduler
from latentsync.models.unet import UNet3DConditionModel
from latentsync.pipelines.lipsync_pipeline import LipsyncPipeline
from diffusers.utils.import_utils import is_xformers_available
from accelerate.utils import set_seed
from latentsync.whisper.audio2feature import Audio2Feature
def main(video_path, audio_path, progress=gr.Progress(track_tqdm=True)):
inference_ckpt_path = "checkpoints/latentsync_unet.pt"
unet_config_path = "configs/unet/second_stage.yaml"
config = OmegaConf.load(unet_config_path)
print(f"Input video path: {video_path}")
print(f"Input audio path: {audio_path}")
print(f"Loaded checkpoint path: {inference_ckpt_path}")
is_shared_ui = True if "fffiloni/LatentSync" in os.environ['SPACE_ID'] else False
temp_dir = None
if is_shared_ui:
temp_dir = tempfile.mkdtemp()
cropped_video_path = process_video(video_path)
print(f"Cropped video saved to: {cropped_video_path}")
video_path=cropped_video_path
trimmed_audio_path = process_audio(audio_path, temp_dir)
print(f"Processed file was stored temporarily at: {trimmed_audio_path}")
audio_path=trimmed_audio_path
scheduler = DDIMScheduler.from_pretrained("configs")
if config.model.cross_attention_dim == 768:
whisper_model_path = "checkpoints/whisper/small.pt"
elif config.model.cross_attention_dim == 384:
whisper_model_path = "checkpoints/whisper/tiny.pt"
else:
raise NotImplementedError("cross_attention_dim must be 768 or 384")
audio_encoder = Audio2Feature(model_path=whisper_model_path, device="cuda", num_frames=config.data.num_frames)
vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16)
vae.config.scaling_factor = 0.18215
vae.config.shift_factor = 0
unet, _ = UNet3DConditionModel.from_pretrained(
OmegaConf.to_container(config.model),
inference_ckpt_path, # load checkpoint
device="cpu",
)
unet = unet.to(dtype=torch.float16)
# set xformers
if is_xformers_available():
unet.enable_xformers_memory_efficient_attention()
pipeline = LipsyncPipeline(
vae=vae,
audio_encoder=audio_encoder,
unet=unet,
scheduler=scheduler,
).to("cuda")
seed = -1
if seed != -1:
set_seed(seed)
else:
torch.seed()
print(f"Initial seed: {torch.initial_seed()}")
unique_id = str(uuid.uuid4())
video_out_path = f"video_out{unique_id}.mp4"
pipeline(
video_path=video_path,
audio_path=audio_path,
video_out_path=video_out_path,
video_mask_path=video_out_path.replace(".mp4", "_mask.mp4"),
num_frames=config.data.num_frames,
num_inference_steps=config.run.inference_steps,
guidance_scale=1.0,
weight_dtype=torch.float16,
width=config.data.resolution,
height=config.data.resolution,
)
if is_shared_ui:
# Clean up the temporary directory
if os.path.exists(temp_dir):
shutil.rmtree(temp_dir)
print(f"Temporary directory {temp_dir} deleted.")
return video_out_path
css="""
div#col-container{
margin: 0 auto;
max-width: 982px;
}
"""
with gr.Blocks(css=css) as demo:
with gr.Column(elem_id="col-container"):
gr.Markdown("# LatentSync: Audio Conditioned Latent Diffusion Models for Lip Sync")
gr.Markdown("LatentSync, an end-to-end lip sync framework based on audio conditioned latent diffusion models without any intermediate motion representation, diverging from previous diffusion-based lip sync methods based on pixel space diffusion or two-stage generation.")
gr.HTML("""
<div style="display:flex;column-gap:4px;">
<a href="https://github.com/bytedance/LatentSync">
<img src='https://img.shields.io/badge/GitHub-Repo-blue'>
</a>
<a href="https://arxiv.org/abs/2412.09262">
<img src='https://img.shields.io/badge/ArXiv-Paper-red'>
</a>
<a href="https://huggingface.co/spaces/fffiloni/LatentSync?duplicate=true">
<img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/duplicate-this-space-sm.svg" alt="Duplicate this Space">
</a>
<a href="https://huggingface.co/fffiloni">
<img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/follow-me-on-HF-sm-dark.svg" alt="Follow me on HF">
</a>
</div>
""")
with gr.Row():
with gr.Column():
video_input = gr.Video(label="Video Control", format="mp4")
audio_input = gr.Audio(label="Audio Input", type="filepath")
submit_btn = gr.Button("Submit")
with gr.Column():
video_result = gr.Video(label="Result")
gr.Examples(
examples = [
["assets/demo1_video.mp4", "assets/demo1_audio.wav"],
["assets/demo2_video.mp4", "assets/demo2_audio.wav"],
["assets/demo3_video.mp4", "assets/demo3_audio.wav"],
],
inputs = [video_input, audio_input]
)
submit_btn.click(
fn = main,
inputs = [video_input, audio_input],
outputs = [video_result]
)
demo.queue().launch(show_api=False, show_error=True, share=True, server_name="0.0.0.0")
LatentSync/configs/audio.yaml 0 → 100644
View file @ 5c023842
audio:
num_mels: 80 # Number of mel-spectrogram channels and local conditioning dimensionality
rescale: true # Whether to rescale audio prior to preprocessing
rescaling_max: 0.9 # Rescaling value
use_lws:
false # Use LWS (https://github.com/Jonathan-LeRoux/lws) for STFT and phase reconstruction
# It"s preferred to set True to use with https://github.com/r9y9/wavenet_vocoder
# Does not work if n_ffit is not multiple of hop_size!!
n_fft: 800 # Extra window size is filled with 0 paddings to match this parameter
hop_size: 200 # For 16000Hz, 200 = 12.5 ms (0.0125 * sample_rate)
win_size: 800 # For 16000Hz, 800 = 50 ms (If None, win_size = n_fft) (0.05 * sample_rate)
sample_rate: 16000 # 16000Hz (corresponding to librispeech) (sox --i <filename>)
frame_shift_ms: null
signal_normalization: true
allow_clipping_in_normalization: true
symmetric_mels: true
max_abs_value: 4.0
preemphasize: true # whether to apply filter
preemphasis: 0.97 # filter coefficient.
min_level_db: -100
ref_level_db: 20
fmin: 55
fmax: 7600
LatentSync/configs/scheduler_config.json 0 → 100644
View file @ 5c023842
{
"_class_name": "DDIMScheduler",
"_diffusers_version": "0.6.0.dev0",
"beta_end": 0.012,
"beta_schedule": "scaled_linear",
"beta_start": 0.00085,
"clip_sample": false,
"num_train_timesteps": 1000,
"set_alpha_to_one": false,
"steps_offset": 1,
"trained_betas": null,
"skip_prk_steps": true
}
LatentSync/configs/syncnet/syncnet_16_latent.yaml 0 → 100644
View file @ 5c023842
model:
audio_encoder: # input (1, 80, 52)
in_channels: 1
block_out_channels: [32, 64, 128, 256, 512, 1024]
downsample_factors: [[2, 1], 2, 2, 2, 2, [2, 3]]
attn_blocks: [0, 0, 0, 0, 0, 0]
dropout: 0.0
visual_encoder: # input (64, 32, 32)
in_channels: 64
block_out_channels: [64, 128, 256, 256, 512, 1024]
downsample_factors: [2, 2, 2, 1, 2, 2]
attn_blocks: [0, 0, 0, 0, 0, 0]
dropout: 0.0
ckpt:
resume_ckpt_path: ""
inference_ckpt_path: ""
save_ckpt_steps: 2500
data:
train_output_dir: output/syncnet
num_val_samples: 1200
batch_size: 120 # 40
num_workers: 11 # 11
latent_space: true
num_frames: 16
resolution: 256
train_fileslist: ""
train_data_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/VoxCeleb2/high_visual_quality/train
val_fileslist: ""
val_data_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/VoxCeleb2/high_visual_quality/val
audio_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/mel_new
lower_half: false
pretrained_audio_model_path: facebook/wav2vec2-large-xlsr-53
audio_sample_rate: 16000
video_fps: 25
optimizer:
lr: 1e-5
max_grad_norm: 1.0
run:
max_train_steps: 10000000
validation_steps: 2500
mixed_precision_training: true
seed: 42
LatentSync/configs/syncnet/syncnet_16_pixel.yaml 0 → 100644
View file @ 5c023842
model:
audio_encoder: # input (1, 80, 52)
in_channels: 1
block_out_channels: [32, 64, 128, 256, 512, 1024, 2048]
downsample_factors: [[2, 1], 2, 2, 1, 2, 2, [2, 3]]
attn_blocks: [0, 0, 0, 0, 0, 0, 0]
dropout: 0.0
visual_encoder: # input (48, 128, 256)
in_channels: 48
block_out_channels: [64, 128, 256, 256, 512, 1024, 2048, 2048]
downsample_factors: [[1, 2], 2, 2, 2, 2, 2, 2, 2]
attn_blocks: [0, 0, 0, 0, 0, 0, 0, 0]
dropout: 0.0
ckpt:
resume_ckpt_path: ""
inference_ckpt_path: checkpoints/latentsync_syncnet.pt
save_ckpt_steps: 2500
data:
train_output_dir: debug/syncnet
num_val_samples: 2048
batch_size: 128 # 128
num_workers: 11 # 11
latent_space: false
num_frames: 16
resolution: 256
train_fileslist: /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/all_data_v6.txt
train_data_dir: ""
val_fileslist: ""
val_data_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/VoxCeleb2/high_visual_quality/val
audio_mel_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/mel_new
lower_half: true
audio_sample_rate: 16000
video_fps: 25
optimizer:
lr: 1e-5
max_grad_norm: 1.0
run:
max_train_steps: 10000000
validation_steps: 2500
mixed_precision_training: true
seed: 42
LatentSync/configs/syncnet/syncnet_25_pixel.yaml 0 → 100644
View file @ 5c023842
model:
audio_encoder: # input (1, 80, 80)
in_channels: 1
block_out_channels: [64, 128, 256, 256, 512, 1024]
downsample_factors: [2, 2, 2, 2, 2, 2]
dropout: 0.0
visual_encoder: # input (75, 128, 256)
in_channels: 75
block_out_channels: [128, 128, 256, 256, 512, 512, 1024, 1024]
downsample_factors: [[1, 2], 2, 2, 2, 2, 2, 2, 2]
dropout: 0.0
ckpt:
resume_ckpt_path: ""
inference_ckpt_path: ""
save_ckpt_steps: 2500
data:
train_output_dir: debug/syncnet
num_val_samples: 2048
batch_size: 64 # 64
num_workers: 11 # 11
latent_space: false
num_frames: 25
resolution: 256
train_fileslist: /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/hdtf_vox_avatars_ads_affine.txt
# /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/hdtf_voxceleb_avatars_affine.txt
train_data_dir: ""
val_fileslist: /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/vox_affine_val.txt
# /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/voxceleb_val.txt
val_data_dir: ""
audio_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/mel
lower_half: true
pretrained_audio_model_path: facebook/wav2vec2-large-xlsr-53
audio_sample_rate: 16000
video_fps: 25
optimizer:
lr: 1e-5
max_grad_norm: 1.0
run:
max_train_steps: 10000000
mixed_precision_training: true
seed: 42
LatentSync/configs/unet/first_stage.yaml 0 → 100644
View file @ 5c023842
data:
syncnet_config_path: configs/syncnet/syncnet_16_pixel.yaml
train_output_dir: debug/unet
train_fileslist: /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/all_data_v6.txt
train_data_dir: ""
audio_embeds_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/whisper_new
audio_mel_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/mel_new
val_video_path: assets/demo1_video.mp4
val_audio_path: assets/demo1_audio.wav
batch_size: 8 # 8
num_workers: 11 # 11
num_frames: 16
resolution: 256
mask: fix_mask
audio_sample_rate: 16000
video_fps: 25
ckpt:
resume_ckpt_path: checkpoints/latentsync_unet.pt
save_ckpt_steps: 5000
run:
pixel_space_supervise: false
use_syncnet: false
sync_loss_weight: 0.05 # 1/283
perceptual_loss_weight: 0.1 # 0.1
recon_loss_weight: 1 # 1
guidance_scale: 1.0 # 1.5 or 1.0
trepa_loss_weight: 10
inference_steps: 20
seed: 1247
use_mixed_noise: true
mixed_noise_alpha: 1 # 1
mixed_precision_training: true
enable_gradient_checkpointing: false
enable_xformers_memory_efficient_attention: true
max_train_steps: 10000000
max_train_epochs: -1
optimizer:
lr: 1e-5
scale_lr: false
max_grad_norm: 1.0
lr_scheduler: constant
lr_warmup_steps: 0
model:
act_fn: silu
add_audio_layer: true
custom_audio_layer: false
audio_condition_method: cross_attn # Choose between [cross_attn, group_norm]
attention_head_dim: 8
block_out_channels: [320, 640, 1280, 1280]
center_input_sample: false
cross_attention_dim: 384
down_block_types:
[
"CrossAttnDownBlock3D",
"CrossAttnDownBlock3D",
"CrossAttnDownBlock3D",
"DownBlock3D",
]
mid_block_type: UNetMidBlock3DCrossAttn
up_block_types:
[
"UpBlock3D",
"CrossAttnUpBlock3D",
"CrossAttnUpBlock3D",
"CrossAttnUpBlock3D",
]
downsample_padding: 1
flip_sin_to_cos: true
freq_shift: 0
in_channels: 13 # 49
layers_per_block: 2
mid_block_scale_factor: 1
norm_eps: 1e-5
norm_num_groups: 32
out_channels: 4 # 16
sample_size: 64
resnet_time_scale_shift: default # Choose between [default, scale_shift]
unet_use_cross_frame_attention: false
unet_use_temporal_attention: false
# Actually we don't use the motion module in the final version of LatentSync
# When we started the project, we used the codebase of AnimateDiff and tried motion module, the results are poor
# We decied to leave the code here for possible future usage
use_motion_module: false
motion_module_resolutions: [1, 2, 4, 8]
motion_module_mid_block: false
motion_module_decoder_only: false
motion_module_type: Vanilla
motion_module_kwargs:
num_attention_heads: 8
num_transformer_block: 1
attention_block_types:
- Temporal_Self
- Temporal_Self
temporal_position_encoding: true
temporal_position_encoding_max_len: 16
temporal_attention_dim_div: 1
zero_initialize: true
LatentSync/configs/unet/second_stage.yaml 0 → 100644
View file @ 5c023842
data:
syncnet_config_path: configs/syncnet/syncnet_16_pixel.yaml
train_output_dir: debug/unet
train_fileslist: /mnt/bn/maliva-gen-ai-v2/chunyu.li/fileslist/all_data_v6.txt
train_data_dir: ""
audio_embeds_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/whisper_new
audio_mel_cache_dir: /mnt/bn/maliva-gen-ai-v2/chunyu.li/audio_cache/mel_new
val_video_path: assets/demo1_video.mp4
val_audio_path: assets/demo1_audio.wav
batch_size: 2 # 8
num_workers: 11 # 11
num_frames: 16
resolution: 256
mask: fix_mask
audio_sample_rate: 16000
video_fps: 25
ckpt:
resume_ckpt_path: checkpoints/latentsync_unet.pt
save_ckpt_steps: 5000
run:
pixel_space_supervise: true
use_syncnet: true
sync_loss_weight: 0.05 # 1/283
perceptual_loss_weight: 0.1 # 0.1
recon_loss_weight: 1 # 1
guidance_scale: 1.0 # 1.5 or 1.0
trepa_loss_weight: 10
inference_steps: 20
seed: 1247
use_mixed_noise: true
mixed_noise_alpha: 1 # 1
mixed_precision_training: true
enable_gradient_checkpointing: false
enable_xformers_memory_efficient_attention: true
max_train_steps: 10000000
max_train_epochs: -1
optimizer:
lr: 1e-5
scale_lr: false
max_grad_norm: 1.0
lr_scheduler: constant
lr_warmup_steps: 0
model:
act_fn: silu
add_audio_layer: true
custom_audio_layer: false
audio_condition_method: cross_attn # Choose between [cross_attn, group_norm]
attention_head_dim: 8
block_out_channels: [320, 640, 1280, 1280]
center_input_sample: false
cross_attention_dim: 384
down_block_types:
[
"CrossAttnDownBlock3D",
"CrossAttnDownBlock3D",
"CrossAttnDownBlock3D",
"DownBlock3D",
]
mid_block_type: UNetMidBlock3DCrossAttn
up_block_types:
[
"UpBlock3D",
"CrossAttnUpBlock3D",
"CrossAttnUpBlock3D",
"CrossAttnUpBlock3D",
]
downsample_padding: 1
flip_sin_to_cos: true
freq_shift: 0
in_channels: 13 # 49
layers_per_block: 2
mid_block_scale_factor: 1
norm_eps: 1e-5
norm_num_groups: 32
out_channels: 4 # 16
sample_size: 64
resnet_time_scale_shift: default # Choose between [default, scale_shift]
unet_use_cross_frame_attention: false
unet_use_temporal_attention: false
# Actually we don't use the motion module in the final version of LatentSync
# When we started the project, we used the codebase of AnimateDiff and tried motion module, the results are poor
# We decied to leave the code here for possible future usage
use_motion_module: false
motion_module_resolutions: [1, 2, 4, 8]
motion_module_mid_block: false
motion_module_decoder_only: false
motion_module_type: Vanilla
motion_module_kwargs:
num_attention_heads: 8
num_transformer_block: 1
attention_block_types:
- Temporal_Self
- Temporal_Self
temporal_position_encoding: true
temporal_position_encoding_max_len: 16
temporal_attention_dim_div: 1
zero_initialize: true
LatentSync/data_processing_pipeline.sh 0 → 100644
View file @ 5c023842
#!/bin/bash
python -m preprocess.data_processing_pipeline \
--total_num_workers 20 \
--per_gpu_num_workers 20 \
--resolution 256 \
--sync_conf_threshold 3 \
--temp_dir temp \
--input_dir /mnt/bn/maliva-gen-ai-v2/chunyu.li/VoxCeleb2/raw
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment