Reputation: 67
I am attempting to implement an audio classifier in my mobile app. When training the data, I used melspectrogram extracted from the raw audio. I am using Tensorflow Lite to integrate the model into the app.
The problem is that I need to perform the same feature extraction on the input audio from the mic before passing into the tflite model. Python's Librosa library implements all of the functions that I need. My initial idea was run Python in flutter (there is the starflut Flutter package but I couldn't get it to work).
Am I going about this in the wrong way? If so, what should I be doing? I could potentially rewrite the Librosa functions in dart lang, but I don't particularly want to do that.
Upvotes: 0
Views: 1121
Reputation: 11807
If you are targeting mobile, check the integration with “native” code. E.g. look for a java/kotlin library that can do the same on android. And a swift/objC one for iOS. Then, you could wrap that functionality in a platform-specific module.
Upvotes: 1