rpvelloso / libjukebox Goto Github PK

to reproduce:

auto sound = jukebox::factory::makeSound(soundFileWithSomeLength);
sound.play();
// sleep or busy wait a little bit, but not too much
sound.play();  // this will be ignored

tested on Windows

=)

you can reproduce it with a small wav file. play it with looping, wait it to be looped and then use .stop() and .play() afterwards

it won't play again even using loop(true)

to gradually stop playing a sound object

expose only the decoder to the soundimpl, remove soundfile from it. This will ease decoration at the decoder for resampling, resolution change, etc.

dr_mp3 has better API

🥇

Now they are reading and decoding the entire files, which can drain the application memory if there are too many objects and start causing memory trashing!

Both wav and ogg files have this behavior.

We need a way to specify a maximum chunk size and design some sort of streaming when playing Sound objects.

https://stackoverflow.com/questions/32374924/simple-reverb-alghoritm-when-buffer-is-small

• move state transitions outside the states, make transitions explicit;
• do not detach playing thread and join it upon state destruction/transition.

Amalgamate source into a single file.

When a soundfile is loaded from a stream, chunk by chunk, it can't be played simultaneously by multiple sound objects (i.e., it can't be shared) because the stream pointer can't be shared (it could be shared using mutex and alike, but we want to avoid this). So a stream soundfile should be allowed to instantiate only one decoder. If sharing is needed then the soundfile should be loaded to memory so it can be played simultaneously by several sound objects (or, if the file is too big, multiples soundfile should be created for the same disk file).

👯‍♂️