Скачать или смотреть Real Time Audio Visualizer as Mel-Spectrogram

Audio classification using Machine learning is done by coverting an audio to an image and thereafter it becomes a classic image classification problem. An audio can be represented as an image in the form a waveform/spectrogram/Mel-spectrogram based on different parameters it is plotted. Mel-Spectrogram is widely for audio classification. In this video, I have created a firefox extesnion leveraging Cursor AI IDE. The extension has start/stop an audio clip. Random audio clips are taken from web of duration 5-10 seconds. A progress bar is to display the remaining time of audio. Live Mel-spectrogram is plotted and you can see different patterns of it with different audio files. Due to this differentiating pattern, machine learning models are trained to classify the audio such as sound of a cat and dog. The classification part is not covered here, but can be looked at https://github.com/praveenraghuvanshi/soun.... Full source is generated by Cursor AI just using some prompts such as below.

"Create a Firefox chrome extension. There should be buttons to start/stop an audio capture. Disable stop button initally. Clicking on start button should pick a free sound clip from web and play it. While sound clip is playing, audio should be transformed to a Mel-Spectrogram and it should be displayed as a preview within the extension. Clicking on stop should stop the audio only. Do not close the Mel-Spectrogram preview window. While clip is playing, disable the Start button. Enable start button once clip is over or stop is clicked. Clicking on start again should clear the previous sound clip and Mel-Spectrogram preview. Also, display the web url of sound clip within the extension. Apply some styling to make the UI look professional."