#
AICoverMaker
Last update: Mar 8, 2025
#
#
Introduction
AICoverMaker is a better and updated version of the old AICoverGen.
This are ideal for users who want 'quick & dirty' AI covers, as the whole process of inputting audio, vocal isolation & song mixing is automated.
# There are two versions this AI Cover maker which are: AICoverMaker Colab and AICoverMaker Kaggle.
#
#
AICoverMaker Colab
#
#
Installation & Setup
1. Go to the AICoverMaker Colab and run the first cell.
Installation may take a couple of minutes, be patient.
2. Next go to the second cell and put your ngrok token in the text box and run it. If you dont have a ngrok acount sign up here and you can authenticate your ngrok tunnel agent here.
Warning
There is a monthly limit rate with Ngrok so dont be supprised if training is suddenly interrupted.
2a. You can choose not to use Ngrok if you desire to by running the third cell.
3. Once you've run either the cell with or without ngrok a link under the cell will apear, click it and it will take you the WebUI.
#
#
Downloading Music & Models
#
Music Download
1. When in the UI look at the top left and look for the tab named Download Music and click it.
2. Then put the song you want to download in the text box and click download.
#
Model Download
1. In the UI look at the top left and look for the tab named Download Model and click it.
2. Then put the model you want to download in the text box and click download.
3. You can also drag and drop your model in the Drop files box to upload them directly.
#
#
AICoverMaker Kaggle
#
#
Installation & Setup
1. Go to the AICoverMaker Kaggle notebook and click Copy & Edit on the top right.
2. Once you're in the notebook run the first installation cell. Once it's done it will output Requirements installed.
Installation may take a couple of minutes, be patient.
3. Next go to the second cell and put your Ngrok token in the TOKEN HERE spot and run it. If you dont have a Ngrok acount sign up here and you can authenticate your ngrok tunnel agent here.
Warning
There is a monthly limit rate with Ngrok so dont be supprised if training is suddenly interrupted.
3. Once you've run the final cell a ngrok link will apear under the cell, click it and it will take you the WebUI.
#
#
Downloading Music & Models
#
Music Download
1. When in the UI look at the top left and look for the tab named Download Music and click it.
2. Then put the song you want to download in the text box and click download.
#
Model Download
1. In the UI look at the top left and look for the tab named Download Model and click it.
2. Then put the model you want to download in the text box and click download.
3. You can also drag and drop your model in the Drop files box to upload them directly.
#
Inference
Please use our Inference Settings guide to find out the inference settings do what.
TTA - results in longer separation time, it gives a little better SDR score but hard to tell if it's really audible in most cases". it “means "test time augmentation", it will do 3 passes on the audio file instead of 1. 1 pass with be with original audio. 1 will be with inverted stereo (L becomes R, R become L). 1 will be with phase inverted and then results are averaged for final output.