Comments (11)
So the dataloader keeps continuing if the training images/audios are absent. Please check if you have provided the right path, and also add print statements in the dataloader to check where it is stuck in an infinite loop.
from wav2lip.
@prajwalkr thanks for the reply have solved it by checking the directory which was not having .jpg inside it. I have 2 questions
- Traning started successfully but It's not saving weights.
- When should I stop training and what is
L1
,Sync
,Percep
,Fake
,Real
by seeing what var should I decide how is the training going
L1: 0.03487106040120125, Sync: 0.0, Percep: 27.63102149963379 | Fake: 0.0, Real: 27.63102149963379: : 1it [00:02, 2.13s/it]
Starting Epoch: 771
L1: 0.03605598956346512, Sync: 0.0, Percep: 27.63102149963379 | Fake: 0.0, Real: 27.63102149963379: : 1it [00:02, 2.12s/it]
Starting Epoch: 772
L1: 0.03456452488899231, Sync: 0.0, Percep: 27.63102149963379 | Fake: 0.0, Real: 27.63102149963379: : 1it [00:02, 2.13s/it]
Starting Epoch: 773
from wav2lip.
why is your training epoch only 1it? The checkpoint saves for every 2000 or 3000 iterations.
from wav2lip.
thanks for the reply its default 1it only where can i change it
from wav2lip.
Are you training on LRS2? If you are training correctly on LRS2, it'll give you several hundreds of iterations per epoch.
from wav2lip.
No, I am Training it on the Lip2Wav
Chemistry Lectures dataset
from wav2lip.
The code is specifically meant for training on LRS2. You will need to understand the code and change multiple parts of the code (including the architecture, maybe) to train on other datasets with different FPS and folder structure.
from wav2lip.
It may be better to resample any dataset you use to 25 fps. The chemistry lecture dataset is 30 fps. Also, do let us know about the results you achieve.
from wav2lip.
Ok I will let you know about the results soon thank you for your time
from wav2lip.
filelist是手动写入吗,我在预处理之后并没有文件内容生成
from wav2lip.
是下载的
from wav2lip.
Related Issues (20)
- How to choose the resize_factor size? Image is always too big to run face detection on GPU. HOT 2
- Is there such a thing as wav2lip-2? HOT 1
- Evaluation problem HOT 1
- Question : apple silicon compatibility ? HOT 4
- Fast question - Resolution upgrading HOT 1
- wav2lip_train training error:Expected more than 1 value per channel when training, got input size torch.Size([1, 512, 1, 1]) HOT 4
- 训练卡住,enumerate(train_data_loader)卡住 HOT 9
- 请教各位高手,有效果类似,但是可以在手机端(ios、Android)运行的对口型模型或者算法吗? HOT 1
- How to stream chatbot text to Wav2Lip on a website so it read out the text in Pyhton?
- Test
- Long video speech synthesis picture jitter(长视频语音合成画面抖动)
- How to remove the box at the mouth of a person(如何去掉人嘴处的方框) HOT 3
- Mic Input for WebUI
- 用预模型请解决下报错 HOT 1
- Can't install requirements HOT 2
- inference face error
- Upper row front teeth are too big, very unnatural, how to deal with it? HOT 2
- Colab Notebook example cannot be run HOT 9
- Outputs of evaluation(LSE-C, LSE-D) consist of four values for each, not one for each? HOT 7
- Mouth position dislocated or some weird artefact's.
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from wav2lip.