This approach has the potential to be extended to merge 2 or more
generated tracks (instruments) with some consideration of volume balance
between the tracks.
One would concatenate the notes in each track, merge and balance the
mp3s and play back the composite mp3 song.
Ron
Ron Wheeler wrote:
If the composer is playing as notes are entered, timing is not an
issue. Fancy electric organ for the typest :-)
If you have a playback button that plays the composed work, you can
concatenate the sounds/notes "selected" and play the composed segment.
Ron
Abe Pazos wrote:
I guess this is meant for Kerry? I think he IS playing songs in a
sequence. Maybe he needs to synchronize graphic elements to the
starting of parts?
In my case, I've built a composer tool and the order is not
predefined, actually is ever changing, so I can't merge the sounds.
Why not just concatenate the mp3 files into 1 file and play it as a
single sound track?
Ron
Abe Pazos wrote:
I don't understand how MP3 can take longer to start. Isn't the AIFF
converted to MP3 anyways by the IDE?
I'm preloading the MP3s from disk. I have generated them myself, so
they are
trimmed until the last sample. Is there something that causes random
delays when starting MP3s?
To test the SOUND_COMPLETE event, I have created a very short
sample (about
1 ms long) which acts as a metronome. I count when it has played 5
times, I play a bass drum. The timing I get is quite random. So I
don't understand how do you get such perfect timing with your
playlist class. Or maybe the problem is that I'm looking for much
more precision than what you needed for the music...
Here is my test code:
var pCounter:int = 5;
var pMetro:Sound = new metronome(); // 1 ms sample to act as a timer
var pChannel:SoundChannel = pMetro.play();
pChannel.addEventListener(Event.SOUND_COMPLETE, onComplete);
var pTime:int = getTimer();
function onComplete(tEvent:Event):void {
pChannel.removeEventListener(Event.SOUND_COMPLETE, onComplete);
pChannel = pMetro.play();
pChannel.addEventListener(Event.SOUND_COMPLETE, onComplete);
trace(getTimer() - pTime);
pTime = getTimer();
if (pCounter-- > 0) {
return;
}
pCounter = 5;
(new BassDrum()).play();
}
These are the times traced:
107 285 48 53 55 104 59 14 42 47 57 34 47 45 45 80 15 43 47 49 47
57 33 46 47 44 84 52 69 24 100 40 72 27 63 64 26 42 44 44 73 20
44 45 47 45 54 38 48 45 57 62 21 43 46 45 49 54 35 45 45 46 67
24 47 45 46 50 53 34 45 46 105 33 45 46 46 78 26 56 23 44 106 55
45 46 45 60 38 42 42 55 58 22 46 45 47 47 48 40 47 44 48 72 18 44
For what I've read, since the sound buffer size in XP is 2048 samples,
I should be getting a stable 46 ms delay.
The problem is there with and without trace().
If I can't fix it I will just use it like it is now. Anyways it
adds a
less-machine more-human(drunk) like touch :)
Thanks for your reply,
Abe
As Jason mentioned, I've been going through the same process for a
very demanding client--one of the country's best music schools. My
project is to string together several audio clips, and
make it sound like a real song,
with
absolutely no glitches. Here's what I've found:
- Use AS3. AS2 isn't fast enough.
- Timers have limited usability, because they can be off by
several milliseconds. I tried that approach, and abandoned it
after the prototype stage.
- MP3's take too long to start playing, and tend to have glitches
like pops and clicks. A program Steven Sacks recommended,
MP3Trim, helped, but still didn't raise it to the
professional level.
- Aiff files at 44.1/16 seem to be the best solution (wav files
would
probably
work too, but my client does all the asset prep on the Mac).
I wrote a playlist class that has methods to build a playlist and
to play
it.
The play method simply steps through the array of sounds you have
built.
It's
its own listener for the SOUND_COMPLETE event, and keeps playing
until it reaches the end of the playlist.
That approach works for all but a few songs, where it's just not
precise enough. For example, there's a pretty hyper guitar solo
from "American
Idiot",
with fast 16th notes (tempo about 128), and I needed to drop in
one
sixteenth
note in the middle of a run. The delay was just long enough--maybe
1/100 second--to make it sound like a hesitation. We had to drop
that song from
the
game.
Cordially,
Kerry Thompson
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
