I want an example where load a wav file and play it, and when it finished sound start again and again and again...

I want to use the function play_audio_stream, but I dont know how.

I try these, but not soud anything:

int main()
{
allegro_init();
install_keyboard();
install_sound(DIGI_AUTODETECT, MIDI_AUTODETECT, NULL);

//create the audio stream
AUDIOSTREAM *stream = play_audio_stream(1024, 8, FALSE, 22050, 255, 128);

//Them create the sample "archivo" loading a wav
SAMPLE *archivo=load_wav("wav.wav");
while(!key[KEY_ESC]){
void *mem_chunk;
mem_chunk = get_audio_stream_buffer(stream);

//when the sound buffer is void:
if (mem_chunk != NULL){

//Put into the buffer the sample "archivo"
mem_chunk=&archivo;

//Indicate that the buffer is ready
free_audio_stream_buffer(stream);
}
}
}

I am from uruguay and my english is basic, please write correctly

I want do it want do it with play_audio_stream

my reason:
"The audio stream functions are for playing digital sounds that are too big to fit in a regular SAMPLE structure, either because they are huge files that you want to load in pieces as the data is required, or because you are doing something clever like generating the waveform on the fly."
http://alleg.sourceforge.net/latestdocs/en/alleg028.html

The line 'mem_chunk = &archivo;' does nothing. You have to fill the buffer using the pointer provided to you by get_audio_stream_buffer.

Example :

If you're going to use SAMPLE*s, then use the routines for playing samples - play_sample.

void* mem_chunk;
    mem_chunk = get_audio_stream_buffer(stream);
    int bytes = length * (bits / 8) * (stereo ? 2 : 1);
    if (mem_chunk) {
       for (int i = 0 ; i < bytes ; ++i) {
           mem_chunk[i] = data[i];

Sorry, I mislead you. Before you can assign values to the array returned by get_audio_stream_buffer, you must cast it to a specific type. Since you are using a 8 bit mono stream, then you want to cast it to an unsigned char*.

This means that you need to convert your data from signed format into unsigned format.

So, before you had this :

char data[1024];
for(int i=0;i<1024;i++) data[i]=sin(i*2*PI/10)*127;//I create the wave, 1 cycle per 10 sampes

But now you should have this :

unsigned char data[1024];
for (int i = 0 ; i < 1024 ; ++i) {
   float raw = sin((float)i*2.0f*M_PI/10.0);
   unsigned char d = (unsigned char)((raw*127.5f) + 127.5f);
   data[i] = d;
}

Edit
Please use <code>code goes here...</code> tags when posting code. It makes it much easier to read.

There are two buffers for the AUDIOSTREAM :

So fill both when you start the stream, and then only fill the other buffer when get_audio_stream_buffer returns non-null.

Not with an AUDIOSTREAM, no. You can make your own 'buffers' and then copy them into your AUDIOSTREAM buffer though.

You could make your own SAMPLE's, modifying the SAMPLE::data array after making a new sample with create_sample. Then you could just use play_sample.

//AUDIOSTREAM *play_audio_stream(int len, int bits, int stereo, int freq, int vol, int pan);
AUDIOSTREAM *stream = play_audio_stream(1024, 8, FALSE, 22050, 255, 128);

The reason it takes so long to hear the sound again is because you are not filling the buffer fast enough. With only 1024 frames in the buffer at 22050 frames per second, you need to refill the buffer (22050/1024 ~= 21) 21 times every second. You can either use a larger number of frames (len) and refill it less often (which will make the sound take longer to change), or you can continue to use a small number of frames and update the sound more often.

I compiled your program and ran it, and there is a noticeable delay when changing sounds.

I think this has to do with Allegro's internal mixer buffer size, and I don't believe that can be changed by the user. Sorry.

play_sample is much faster for me, and plays pretty much immediately after a key press for me.