AI with Aimee: Voice to Voice AI for Singing Using Google Colab and RVC

In my previous article, we worked from the beginning together all the way to having a fully trained voice model using RVC technology. Here is where the real magic happens! Now we will take the trained voice model and make them sing for us!

As a brief recap, here is the voice model source material. It’s my friend, B3AM:

And this is a vocal sample I licensed from LANDR that we will be using in this tutorial example:

Step One: Let’s get started by connecting to our Google Colab workbook.

Step Two: Click on the run arrow under the Install to Google Drive section.

Step Three: Follow the prompts to connect your Google drive account to the shared Google Colab workspace.

Step Four: Navigate to the audios section in the project-main folder and right-click to select the upload option. Here you will upload the voice inference file you wish your voice model to speak or sing.

Step Five: Copy the path of the voice inference file you uploaded.

Step Six: Complete filling out the inference section, pasting your inference audio path into the “input_path” field. Additional instructions provided below for this section.

  1. In the transpose field, use negative integers to lower the pitch of your file conversion and positive integers to raise the pitch of your file conversion. A value of -12 will lower the pitch by one octave while a value of 12 will raise the pitch by one octave.
  2. Input_path is where you paste the path to the audio file you want your voice model to sing or speak.
  3. Opt_path is not required to modify but you can change the name of the output file based on your preferences here.
  4. Under model_name make sure to list the exact name of the model you trained previously.
  5. After you’ve completed steps 1-4, click on the run arrow to process the output file.
  6. A preview of the output file will appear once it is done being processed. If it’s satisfactory, retrieve it from your Google Drive at the opt_path location you specified in step 3.

Here is the completed setup for this tutorial’s inference section:

And here is the resulting audio file:

If you completed all of the steps above, congratulations! You’ve successfully given your trained voice model a script to say or sing – and hopefully it did! You made it happen, captain!

Until next time! Happy prompting! ^_^

Share the Post:

Related Posts

Join Our Newsletter

AI with Aimee: Voice to Voice AI for Singing Using Google Colab and RVC

In my previous article, we worked from the beginning together all the way to having a fully trained voice model using RVC technology. Here is where the real magic happens! Now we will take the trained voice model and make them sing for us!

As a brief recap, here is the voice model source material. It’s my friend, B3AM:

And this is a vocal sample I licensed from LANDR that we will be using in this tutorial example:

Step One: Let’s get started by connecting to our Google Colab workbook.

Step Two: Click on the run arrow under the Install to Google Drive section.

Step Three: Follow the prompts to connect your Google drive account to the shared Google Colab workspace.

Step Four: Navigate to the audios section in the project-main folder and right-click to select the upload option. Here you will upload the voice inference file you wish your voice model to speak or sing.

Step Five: Copy the path of the voice inference file you uploaded.

Step Six: Complete filling out the inference section, pasting your inference audio path into the “input_path” field. Additional instructions provided below for this section.

  1. In the transpose field, use negative integers to lower the pitch of your file conversion and positive integers to raise the pitch of your file conversion. A value of -12 will lower the pitch by one octave while a value of 12 will raise the pitch by one octave.
  2. Input_path is where you paste the path to the audio file you want your voice model to sing or speak.
  3. Opt_path is not required to modify but you can change the name of the output file based on your preferences here.
  4. Under model_name make sure to list the exact name of the model you trained previously.
  5. After you’ve completed steps 1-4, click on the run arrow to process the output file.
  6. A preview of the output file will appear once it is done being processed. If it’s satisfactory, retrieve it from your Google Drive at the opt_path location you specified in step 3.

Here is the completed setup for this tutorial’s inference section:

And here is the resulting audio file:

If you completed all of the steps above, congratulations! You’ve successfully given your trained voice model a script to say or sing – and hopefully it did! You made it happen, captain!

Until next time! Happy prompting! ^_^

Share the Post:

Related Posts