Introducing Stable Audio Open: your new go-to for effortless sound generation! Bid farewell to costly sound libraries and complicated audio production software. With this cutting-edge AI model, you can generate any sound with just one click. From drum beats and instrument riffs to ambient sounds and sound effects, this revolutionary tool creates up to 47 seconds of high-quality audio from a simple text prompt. In this tutorial, I'll demonstrate how to quickly generate the perfect sound for your project. Ideal for music producers, sound designers, and anyone aiming to elevate their audio projects with ease!
To start off let's make sure we have all the requirements to begin. Here’s what you need.
To run Stable Audio Tools you need to have Python installed on your computer.
You need Git for Windows to easily download software or tools directly from repositories like GitHub.
To begin, create an empty folder on your hard drive.
Next we need to paste a few more lines in the same command prompt to install the cuda version of torch.
Once Torch is installed, the final step is to install the packages using “pip install .”
While the packages are being installed we can go ahead and download the stable-audio-open-1.0 model from Hugging Face. You need to create an account and fill out the information to access this model.
Once you’ve done that go and head to “Files and versions” and download “model.safetensors” and “model_config.json”.
Once the model and the config have been downloaded we will move them inside a new folder that we have to create inside the stable-audio-tools folder. Create a new folder named “ckpt” and move the safetensors and json file in here.
We can now run the WebUI by running the following command in the command prompt.
python run_gradio.py --ckpt-path ".\ckpt\model.safetensors" --model-config ".\ckpt\model_config.json"
Now you should see a local URL, hold CTRL and click on the link to launch the WebUI!
With the WebUI launched we can finally have some fun and start generating some awesome sound effects!
I recommend starting at 100 steps to test your prompts and when you’re satisfied increase the steps to 300-400.
With the settings adjusted hit “Generate” and listen as AI creates amazing sounds for you! It’s that easy!
As you may have noticed there is no .bat file to launch the stable-audio-tools WebUI. Luckily it’s very easy to create one ourselves. Head to the stable-audio-tools root folder and create a new text document, name it “start_webui”. Open the text file and paste the following:
@echo off python -m venv env timeout /t 2 /nobreak > nul call env\Scripts\activate timeout /t 1 /nobreak > nul python run_gradio.py --ckpt-path ".\ckpt\model.safetensors" --model-config ".\ckpt\model_config.json"
Now we will save it as a .bat file that we can run to launch the WebUI.
Now when you want to run Stable Audio Tools simply double click the “start_webui.bat” file. This will launch the command prompt and will create the local URL for you.
Congratulations! You've now mastered the use of Stable Audio Open to effortlessly generate high-quality sound effects and music. By following this tutorial, you've set up the necessary tools, installed the required software, and learned how to operate the Stable Audio Tools WebUI to create custom audio with just a few clicks. Keep experimenting with different prompts and settings to explore the full potential of this innovative tool. Happy sound generating!
Stable Audio Open is an AI-powered tool that generates high-quality sound effects and music from simple text prompts, ideal for music producers and sound designers.
Install Python and Git, then clone the Stable Audio Tools repository. Follow the setup instructions to create a virtual environment and install necessary packages
Stable Audio Open simplifies the process of creating custom audio, saving time and money on costly sound libraries and complex production software.