0

I am a back-office system developer, and am struggling to deal with a problem.

I am addicted to podcasts (in fact I use the Podcast Addict app), but I am interested in all kinds of subjects and there are a lot of podcasters to provide me a bunch of options. This addiction means that I have more podcasts to listen to than hours in a week. So I listen with accelerated audio.

But, podcast addicted implementation of accelerated audio is very power hungry, and I am wondering if it should.

My goal is to get any information about CPU/GPU usage of every podcast player when playing in accelerated playback.

Any advice? Tools? Procedures? Anyone has made this analysis before?

| improve this question | |
0

Looking into the APK of Podcast Addict you can find the library libsonic which is the key component "for speeding up or slowing down speech". As it is a native library it should be relative efficient (compared to plain Java). The sonic library does not seem to use the GPU for computations, just plain CPU power.

Regarding the comparison on different Apps modern Android systems already have everything in place to do so. On the app details page you can see the CPU time (total, background), and the keep wake time.

If you want to see more details you can use an app like GSam Battery Monitor.

Using a fixed length podcast you can compare different apps by noting the displayed values before and after the test.

For testing non-owned apps I only see the chance to buy it if it is not free, test it and then if it does not meet your expectations regarding the battery usage refund it. On Google Play Store this should be possible within 48 hours after you have bought the app.

| improve this answer | |
  • Thanks a lot. But by 'not owned' I meant that I am not the developer, so I think that it would be impossible to make any profiling (am I wrong?). I was thinking of connecting to Android studio for any deeper diagnostic. To me seems vary strange that it does not use GPU, I mean, the acceleration (being done in real-time) is a algebraic wave compression, GPU should excel doing it, and would be battery conservative as well. I am totally Ok with buying a few apps for testing purposes, but, would be better to have a nice setup (profiling) to test them quickly. – Rafareino Jul 21 '18 at 11:08
  • That you don't have the source does not make a difference as long as you don't want to debug the app (as debugging slows it down this is nothing you want to do). I think the library is too old for OpenCL. Also using the GPU means you have to re-invent your existing wheel, as you can't reuse the code and just run it on the GPU. – Robert Jul 21 '18 at 11:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.