There are two levels of sample playing functions. You can mix and match them, but it's important to understand the differences.

1. Higher Level: al_reserve_samples(int n) is paired with al_play_sample(). n how many times you can simultaneously call that latter function, regardless if it's the same underlying raw data or not.

   It creates its own voice/mixer combination if you don't tell it about a default one via al_set_default_mixer(). So in your example, you should set the default mixer before calling the reserve sample function.

   
   

2. Lower Level: create voice / create mixer / attach x to y / al_create_sample_instance() / al_play_sample_instance() If you want to play the same sample N simultaneous times, then you need to create 1 ALLEGRO_SAMPLE and n ALLEGRO_SAMPLE_INSTANCE variables.

The two structures:

• ALLEGRO_SAMPLE: the raw, uncompressed data.

  
  

• ALLEGRO_SAMPLE_INSTANCE: a playing instance of that data; multiple of these can exist per ALLEGRO_SAMPLE. When using al_play_sample(), you don't have access to the ALLEGRO_SAMPLE_INSTANCE, because Allegro manages that for you.

So in your case, you may want to move toward this (with the lower level API):

typedef struct SFX {
   ALLEGRO_SAMPLE *samp;
   ALLEGRO_SAMPLE_INSTANCE **inst;
   int max_simultaneous_count;
} SFX;

When loading the sample, set max_simultaneous_count and create an array of that many ALLEGRO_SAMPLE_INSTANCE. Then when playing your SFX, loop through that array and play the sample instance if it is not currently playing.

Okay, that clarifies things. I still can't adjust the sound's speed while it is playing.

Anyhow, I removed this code from my game for now and went back to using al_play_sample() (thank God for backups). All this headache just isn't worth it for one small sound effect.

I'll have to figure t his out another day, I may take a peek at the al_play_sample() code sometime and see what it does if I can understand it, but if I spend much more time trying to figure this out I will never get this game done, and I just want to finish this thing sometime this century.

It does look like, if I ever plan to use this that I will have to write lots of code to track multiple instances in some sort of array, and I somehow track when they're done playing yada yada yada... I am starting to miss Allegro 4, at least for audio anyhow.

I love Allegro 5, but not the audio system, seems overly complicated compared to Allegro 4.

Edit: Thanks for the additional ideas. I will still work on this on the side, in a separate project so I can continue to try and get something working that I can use in all my future projects. It's just this particular project has dragged on too long now and I tire of working on it (but refuse to stop until it is done! ).

Edit2:

Good idea, but I am wondering. Do you have to check if that instance of a sample is still playing? When it ends playing, do you destroy that instance? I'm just uncertain about the management of all of this.