Comments (17)
@agupta54 with --streaming
I think the problem should go away.
from denoiser.
How much RAM do you have? Can you try passing --num_workers=1
to the command?
from denoiser.
I also had this issue and --streaming
flag solved it.
Thank you @adefossez for the tip!
from denoiser.
@adefossez I am using Google Colab GPU which gives me 12 GB RAM
I tried to pass --num_workers=1 but it gives me that and it didn't denoise the audio
from denoiser.
How big is your file ? You could also try to pass --streaming
, although I'm not sure it would help.
from denoiser.
@adefossez it is 15 mins wav file, i will try it
from denoiser.
@adefossez I splitted the audio into (3sec) audios and the code run but the result was horrible!
https://drive.google.com/file/d/1-0zOzQ2Fo0f3qwxnmLZ6Abnqc2bOoqSI/view?usp=sharing
from denoiser.
3 seconds is really too short, you should do at least 10 secs. have you tried the streaming version ? normally I would expect it to use very little memory overall.
from denoiser.
@adefossez I tried the streaming version, I passed a 10 mins audio and the output is so BAD
https://drive.google.com/file/d/1-EOc4XVHvK919U4xprsj9T8_XIcJY1Rr/view?usp=sharing
from denoiser.
is your audio sampled at 16khz and mono channels ?
from denoiser.
@adefossez no its sampling rate is 44100 and i am passing this at the command and yes it is mono it was stereo but i converted it into mono also its extension was mp3 and i converted in into wav
from denoiser.
the audio must be 16kHz :)
from denoiser.
@adefossez I tried other audios with sampling rate otherthan 16KHZ and passed the --sampling_rate value in the command and it worked, i think the problem with conversion of stereo into mono in this audio file
from denoiser.
Well if you pass the --sampling_rate
value it is not going to adapt what the model has been trained for, so you still need to resample to 16khz
from denoiser.
@adefossez so what is the use of --sampling_rate
from denoiser.
The use is to specify the sample rate you trained the model with. If you train a new model with a different sample rate then you should use it.
from denoiser.
I am getting the same error when I am trying to enhance files of greater than 23 minutes duration. My files are mono channel, 16 bits per sample and a sample rate of 16000. With --num-workers 1 also I am getting the same error.
from denoiser.
Related Issues (20)
- Help please
- Very distorted output
- Causal model results on Valentini
- My valid loss =0, is this normal? HOT 2
- where are the models?
- Wasm model conversion
- How do I use the pre-train model?
- Question about the implemented of SpectralConvergengeLoss
- Using denoiser at all doesn't work at all HOT 1
- access to pretrained weight HOT 1
- Fine-tuning with custom data
- Data DNS load is bug
- Denoiser doesn't works properly
- Background noise persistant even after running the denoiser
- Commercial use? HOT 2
- Can not remove noise at the first seconds
- Seeking Clarification on DemucsStreamer Logic HOT 1
- how to use local model ?
- Denoise an audio array (ndarray) instead of an audio file (mp3)
- Output with distortion
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 denoiser.