Comments (3)
Hey there, the 24khz model uses 32 codebooks so that we could directly compare to EnCodec. The configuration in terms of stride, number of codebooks, and sampling rate all match EnCodec. You're right though, that further compression should be possible!
However, quantizer dropout was used for this model as well. Here are the quantitative results at each bitrate:
from descript-audio-codec.
I see, thank you for the answer!
Are there any plans to release a 24kHz model trained with fewer codebooks as well?
I'm not sure I understand the concept of these residual vector quantized codecs yet, so this might be a stupid question, but to reduce the bitrate of the model you just use fewer codebooks, right? So if the 24kHz model uses 32 codebooks to have a 24kbps bitrate, using only 8 codebooks corresponds to 6kbps?
from descript-audio-codec.
@Flux: yes, 8 codebooks corresponds to 6kbps.
from descript-audio-codec.
Related Issues (20)
- Error when set win_duration small
- Encoding new file - use of `zero_pad` HOT 2
- Inference speed
- Loading DAC files is insecure due to pickle
- Error with 16khz
- Memory leak? HOT 2
- Padding Mismatches Output Dimension in Conv1d HOT 1
- broken training: please specify versions of libraries used
- tensor shape mismatch when training on 24khz LibriTTS dataset HOT 2
- Same error in #18
- Very low bitrate models
- Training error: "RuntimeError: grad can be implicitly created only for scalar outputs"
- How to directly download the trianing data of baseline?
- Decode using codes instead of encoder output? HOT 2
- Duration not preserved HOT 1
- Duration not preserved?
- Strange at the end of the recons audio
- The size of tensor a (5) must match the size of tensor b (6) at non-singleton dimension 1 HOT 1
- remove weight norm when inference?
- Docs on Model Evaluation pipeline
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.