Using tensorflow to specify sounds

MIT App Inventor Help
1

Hi
So i want to be able to use a tensorflow model, is it basically possible?

And i also want to convert input sound into a spectrogram

Are these possible in mit app inventor

2

When I search for „ tensorflow model,“ I get 22 results. Did you read them? Before posting first step should be a search in this community!

3

i read some of them but idk if it means that i can use tensorflow or not

4

Maybe you should read all of them.
here is what Gemini says about your question:
Yes, it is possible to use TensorFlow to specify sounds in App Inventor.

The MIT App Inventor platform has a dedicated extension called the "Personal Audio Classifier" (PAC) that allows users to train and deploy custom audio classification models. This tool utilizes TensorFlow.js under the hood, enabling you to train models in your browser and then integrate them into your App Inventor projects.

Here's how it generally works:

  1. Train your model: You use a web-based interface like the Personal Audio Classifier to train a custom sound classification model. You provide audio samples for different categories (e.g., dog bark, car horn, human speech) and label them. The tool then uses this data to train a TensorFlow.js model.
  2. Export the model: Once trained, you can export the model in a format compatible with App Inventor (often a .mdl file, which contains the TensorFlow Lite model).
  3. Integrate into App Inventor: You import the exported model into your App Inventor project using the Personal Audio Classifier extension. This extension provides blocks that allow your app to record audio, send it to the loaded model for classification, and receive the prediction results.

This allows you to create App Inventor applications that can recognize and respond to specific sounds based on your trained TensorFlow model.

Build a Deep Audio Classifier with Python and Tensorflow - This video demonstrates how to build a deep audio classifier using Python and TensorFlow, which is a foundational step for creating models that can be integrated with App Inventor.