Playing a sequence of sounds without gaps (iPhone) - ios

Question

I thought maybe the fastest way was to go with Sound Services. It is quite efficient, but I need to play sounds in a sequence, not overlapped. Therefore I used a callback method to check when the sound has finished. This cycle produces around 0.3 seconds in lag. I know this sounds very strict, but it is basically the main axis of the program.
EDIT: I now tried using AVAudioPlayer, but I can't play sounds in a sequence without using audioPlayerDidFinishPlaying since that would put me in the same situation as with the callback method of SoundServices.
EDIT2: I think that if I could somehow get to join the parts of the sounds I want to play into a large file, I could get the whole audio file to sound continuously.
EDIT3: I thought this would work, but the audio overlaps:
waitTime = player.deviceCurrentTime;
for (int k = 0; k < [colores count]; k++)
{
player.currentTime = 0;
[player playAtTime:waitTime];
waitTime += player.duration;
}
Thanks

Answer 1

I just tried a technique that I think will work well for you. Build an audio file with your sounds concatenated. Then build some meta data about your sounds like this:
#property (strong, nonatomic) NSMutableDictionary *soundData;
#synthesize soundData=_soundData;
- (void)viewDidLoad {
[super viewDidLoad];
_soundData = [NSMutableDictionary dictionary];
NSArray *sound = [NSArray arrayWithObjects:[NSNumber numberWithFloat:5.0], [NSNumber numberWithFloat:0.5], nil];
[self.soundData setValue:sound forKey:#"soundA"];
sound = [NSArray arrayWithObjects:[NSNumber numberWithFloat:6.0], [NSNumber numberWithFloat:0.5], nil];
[self.soundData setValue:sound forKey:#"soundB"];
sound = [NSArray arrayWithObjects:[NSNumber numberWithFloat:7.0], [NSNumber numberWithFloat:0.5], nil];
[self.soundData setValue:sound forKey:#"soundC"];
}
The first number is the offset of the sound in the file, the second is the duration. Then get your player ready to play like this...
- (void)viewDidAppear:(BOOL)animated {
[super viewDidAppear:animated];
NSURL *url = [NSURL fileURLWithPath:[NSString stringWithFormat:#"%#/audiofile.mp3", [[NSBundle mainBundle] resourcePath]]];
NSError *error;
audioPlayer = [[AVAudioPlayer alloc] initWithContentsOfURL:url error:&error];
audioPlayer.numberOfLoops = -1;
if (audioPlayer == nil)
NSLog(#"%#", [error description]);
else {
[audioPlayer prepareToPlay];
}
}
Then you can build a low-level sound playing method like this ...
- (void)playSound:(NSString *)name withCompletion:(void (^)(void))completion {
NSArray *sound = [self.soundData valueForKey:name];
if (!sound) return;
NSTimeInterval offset = [[sound objectAtIndex:0] floatValue];
NSTimeInterval duration = [[sound objectAtIndex:1] floatValue];
audioPlayer.currentTime = offset;
[audioPlayer play];
dispatch_after(dispatch_time(DISPATCH_TIME_NOW, duration * NSEC_PER_SEC), dispatch_get_current_queue(), ^{
[audioPlayer pause];
completion();
});
}
And you can play sounds in rapid combination like this ...
- (IBAction)playAB:(id)sender {
[self playSound:#"soundA" withCompletion:^{
[self playSound:#"soundB" withCompletion:^{}];
}];
}
Rather than nesting blocks, you could build a higher-level method that takes a list of sound names and plays them one after the other, that would look like this:
- (void)playSoundList:(NSArray *)soundNames withCompletion:(void (^)(void))completion {
if (![soundNames count]) return completion();
NSString *firstSound = [soundNames objectAtIndex:0];
NSRange remainingRange = NSMakeRange(1, [soundNames count]-1);
NSArray *remainingSounds = [soundNames subarrayWithRange:remainingRange];
[self playSound:firstSound withCompletion:^{
[self playSoundList:remainingSounds withCompletion:completion];
}];
}
Call it like this...
NSArray *list = [NSArray arrayWithObjects:#"soundB", #"soundC", #"soundA", nil];
[self playSoundList:list withCompletion:^{ NSLog(#"done"); }];

Answer 2

I'm assuming you want to change the sequence or omit sounds sometimes. (Otherwise you would just build the asset with all three sounds in a row and play that).
There might be a better idea out there, but to get things very tight, you could consider producing that concatenated asset, pre-loading it - moving up all the latency to that one load, then seeking around it to change the sound.