Trackers and metadata
category: general [glöplog]
Greetings all, I've been trying to find a good tracker to help in the music aspect of my demo im working on. I tried a trial copy of Sony Sound Forge but I wasnt pleased by the fact that the metadata was saved to an external file. I'm pretty new to this so, what trackers are available that actually encode the metadata "inside" the music track.
As a side note (which may not be relavent) I'm primarily interested in the Ogg Vorbis format.
As a side note (which may not be relavent) I'm primarily interested in the Ogg Vorbis format.
neither sonic sound forge (or has it really been bought by sony?) nor the ogg format have ANYTHING to do with tracking.
take this for example.
take this for example.
Thank you for the link, let me rephrase my question. How is certain sync information saved within a song file???
Sync Information? You mean the pattern data?
ogg vorbis metadata fields don't store syncing information.
Sound forges saves peak data in a separate file to save time analyzing the waveform everytime you load it.
If you need such information, it won't be encoded in the ogg format directly.
No tracker will do that for you neither.
You may try to hunt for a multiplexer wich may allow you to encode info in the ogg stream. You'd have to write a demultiplexer though, unless your multiplexer comes with a library.
Sound forges saves peak data in a separate file to save time analyzing the waveform everytime you load it.
If you need such information, it won't be encoded in the ogg format directly.
No tracker will do that for you neither.
You may try to hunt for a multiplexer wich may allow you to encode info in the ogg stream. You'd have to write a demultiplexer though, unless your multiplexer comes with a library.
iirc there is special sony mp3 encoder for soundforge that gives you the metadata too..
anyway, the way i do it is that i render it as a wav-file so i get the metadata/markers and then i have a program that export the times+comments of the into an array, works rather nice.
anyway, the way i do it is that i render it as a wav-file so i get the metadata/markers and then i have a program that export the times+comments of the into an array, works rather nice.
Ok, bad choice of words on my part. When I said sync information what meant where tags that I place at specific points in the song. I've done this before and was able to read the tags from my program which I then used to sync what I was rendering with the song. The draw back to that was that the tag data was stored in a seperate file. I want to circumvent the seperate file issue and somehow have everything in the song file itself.
i don't know what you people call "trackers", but ok...
there are some unused effects in s3m, xm and it. i'm sure other formats will have them too - though .rns from renoise (not reason!) will prolly come out in xml sometime soon (or did it come already?) but so far it's still tricky to include rns in a demo.
so. get openMPT, start tracking away, and then you can include some effect data which will not be interpreted as "music" but simple trigger for whatever you wanna do. you can have several effects and several data triggering that.
i remember doing something like that with Tournesol, one year ago. when i thought i was going to be hired by 7th cube :)
there are some unused effects in s3m, xm and it. i'm sure other formats will have them too - though .rns from renoise (not reason!) will prolly come out in xml sometime soon (or did it come already?) but so far it's still tricky to include rns in a demo.
so. get openMPT, start tracking away, and then you can include some effect data which will not be interpreted as "music" but simple trigger for whatever you wanna do. you can have several effects and several data triggering that.
i remember doing something like that with Tournesol, one year ago. when i thought i was going to be hired by 7th cube :)
Couldn't you just abuse the ID3 tag?
Yeah, tracker is definitely the wrong terminology here if you're talking about streamed music.
I only know of .wav having sync markers. I've used Soundforge to add sync markers (for use with FMOD) and then saved as wav using an mp3 codec, I'm not sure if there's a Vorbis codec for wav. Only problem is that there can be slight differences in marker placement in the compressed wav compared to the uncompressed file.
It's definitely easier to go with a tracked format if you want tight sync.
I only know of .wav having sync markers. I've used Soundforge to add sync markers (for use with FMOD) and then saved as wav using an mp3 codec, I'm not sure if there's a Vorbis codec for wav. Only problem is that there can be slight differences in marker placement in the compressed wav compared to the uncompressed file.
It's definitely easier to go with a tracked format if you want tight sync.
You listen to the music, check the at what time events happen then your coder should do something like
Yes. Coders are THAT smart.
Code:
if(time>=EVENT1 && time<=EVENT2)
dothing();
(...)
Yes. Coders are THAT smart.
this is pretty much what I do (read: ugly ass shit piece of code):
NOTE: you need to save as wav file or else those riff "markers" get wiped... so dump your syncpoints from a .wav file... then load an ogg version of it in your demo.
In your demo you read those syncpoints and call a bunch of Fmod apis to setup everything for you:
FSOUND_Stream_SetSyncCallback(hStream,Callback,0);.
FSOUND_Stream_AddSyncPoint(hStream, pcm, name);
hope it helps
Code:
// proggy to dump syncpoints to a file
#include "stdafx.h"
#include "fmod.h"
FSOUND_STREAM * hStream = NULL;
#define MAX_SYNC_POINTS 4096
typedef struct tagSyncPoint {
char name[64];
unsigned int offset;
} SyncPoint;
FSOUND_SYNCPOINT *points;
SyncPoint syncpoints[MAX_SYNC_POINTS];
int _tmain(int argc, _TCHAR* argv[])
{
printf("WAV Marker to FMOD SyncPoint translator v1.0 (c) pnx/qop\r\n");
if(argc<1) {
printf("Usage: m2f.exe filename.wav\r\n");
exit(0);
}
memset(&syncpoints, 0, sizeof(syncpoints));
FSOUND_Init(44100, 2, FSOUND_INIT_GLOBALFOCUS);
hStream = FSOUND_Stream_Open(argv[1], FSOUND_MPEGACCURATE | FSOUND_LOOP_OFF, 0, 0);
if(!hStream) {
printf("file %s not found!\r\n", argv[1]);
exit(0);
}
int i=0;
do {
points = FSOUND_Stream_GetSyncPoint(hStream, i+1);
if(points) {
strncpy(syncpoints[i].name, FSOUND_Stream_GetSyncPointInfo(points, &syncpoints[i].offset), 63);
i++;
}
else
break;
} while(i<MAX_SYNC_POINTS);
printf("found %d syncpoints in the wav file\r\n", i);
for(int j=0;j<i;j++)
printf("sync%d: %s - sample: %u\r\n", j, syncpoints[j].name, syncpoints[j].offset);
FILE *output = fopen("syncpoints.ess", "wb");
fwrite(&i, sizeof(unsigned int), 1, output);
fwrite(syncpoints, sizeof(SyncPoint), i, output);
fclose(output);
FSOUND_Stream_Close(hStream);
FSOUND_Close();
return 0;
}
NOTE: you need to save as wav file or else those riff "markers" get wiped... so dump your syncpoints from a .wav file... then load an ogg version of it in your demo.
In your demo you read those syncpoints and call a bunch of Fmod apis to setup everything for you:
FSOUND_Stream_SetSyncCallback(hStream,Callback,0);.
FSOUND_Stream_AddSyncPoint(hStream, pcm, name);
hope it helps
sidenote: panoramix's stuff is all about fmod 3.
other than that, real men make synctools because some "sync points" just can't pull the trick.
that, or they hit the space bar (ask melwyn)
other than that, real men make synctools because some "sync points" just can't pull the trick.
that, or they hit the space bar (ask melwyn)
panoramix thanks for the example code. Before i switched to the ogg vorbis format I was using mp3's and I was able to track the sync points while the song was playing and I would call my sync callback function (I too am using fmod). I'll do some experimenting to see what results I get. Thank you to all of you.
Code:
if (songpos <= 123400)
stage = 0;
else if (songpos <= 127200)
stage = 1;
else if (songpos <= 131000)
stage = 2;
else if (songpos <= 134800)
stage = 3;
else
stage = 4;
Beat detection \o/
