/spectrogram-android

Primary LanguageJavaOtherNOASSERTION

Synopsis

[Spectrogram-Android] is a mobile application available on the Google Play store.
It was initially developed as a basic tool to study outdoor sound properties.
It is and will remain free to use and ad-free.

Key features
  • Realtime time view and frequency view of surrouding sound
  • FFT size from 32 to 65536
  • Nine windowing functions
  • Night/day and portrait/landscape modes
Snapshots

Installation and configuration

Installation
  • Install Android Studio 2.3.2, Android SDK and Android NDK
  • From the "Android SDK Manager", install the libraries
    • "Android Support Library (Obsolete)"
    • "Android Support Repository"

Congratulations, you are ready to run the application!

Implementation

The application follows Android guidelines in terms of Activity lifecycle, View inflation and Preferences management.
It uses native code (C code) to reduce FFT computation time. A JNI bridge defines the binding between Java methods and C functions.
The code is very minimalist. Below the exhaustive list of source files.

Edit:
Note that the C code and JNI bridge have been moved into the dedicated repository sound-engine-android. They are now packaged into .so files and a .jar file attached to this project.

Java side code
  • SpectrogramActivity.java (entry point of the application, renders TimeView and FrequencyView, has access to native code to process audio samples, subscribes to audio recording service)
  • PreferencesActivity.java (simple interface to manage user preferences)
  • TimeView.java (wave signal rendering, shown on top of the application)
  • FrequencyView.java (spectrogram rendering, shown below the wave signal)
  • ContinuousRecord.java (background service, continuously records audio, calls a callback for each sample recorded)
  • Misc.java (various function to ease preferences manipulation)
Native side code
  • SoundEngine.java (JNI bridge in Java, list of Java methods that are implemented in C)
  • jni.c (JNI bridge in C, allows Java calls to be routed to C functions)
  • main.c (FFT, windowing ... all sound processing functions)

Basically, an instance of ContinuousRecord is run at startup and continuously records audio samples from the microphone.
Using a callback, each sample is processed within SpectrogramActivity. Results are then sent to TimeView and FrequencyView which update their respective content.

Development

Have killer features in mind you want to implement?
You are very welcome to contribute to the project!

Todo

  • Add extra languages
  • Add frequency range filter
  • Add unit test cases (yeah I know...)
  • Add ability to play sound from the spectrogram
  • Add high resolution spectrogram export feature

License

zlib