Comments (3)
We did not use these losses during training. Only multi-scale mel loss and the discriminators were needed. When the waveform loss weight is very low, we found it didn't really do much, and when it was set very high, we found it resulted in buzzy or fuzzy output.
Similarly, multi-scale STFT loss didn't have much improvement.
Something to note is that both of these losses are sorta baked into the discriminators, as there is a feature matching loss. The feature matching loss on the early layers of the discriminators inherently captures waveform and STFT loss.
from descript-audio-codec.
Hi @Aria-K-Alethia: did you manage to start training mechanically? If yes, could please your environment?
from descript-audio-codec.
Got it.
Thank you for your kind reply.
I will then close this issue.
from descript-audio-codec.
Related Issues (20)
- hours of audio content for the training? HOT 1
- can descript-audio-codec works below 32kbps outperforms opus at 32kbps? HOT 4
- entropy computation script not working (it used to work before) HOT 4
- Why does the 24kHz model use 32 codebooks? HOT 3
- How is stereo compression done? HOT 1
- Some question about official log
- Details on training and inferencing the 16kHz variant HOT 2
- finding out the stride of the model HOT 1
- Question regarding kernel size and stride direction in the MRD module HOT 5
- byte count on 16kHz decoding HOT 4
- Which win_duration to use?
- Chunked inference result depends on chunk length HOT 9
- Can't set n_quantizers for encode() HOT 2
- Fine-tuning from 44.1Khz
- 16kHz configs result in shape mismatch HOT 1
- does commit_loss and codebook_loss always be equal? HOT 2
- Different code sizes when encoding versus when compressing HOT 2
- audiotools error
- How to compress stereo sound by model.encode HOT 2
- (Paper Error?) MSD Not Used? HOT 3
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 descript-audio-codec.