[Csnd] sample-accurate mechanism

The Csound Mailinglist
1

Hello everyone!

I try to get my head around some plugin development and i have some questions regarding the sample-accurate mechanism.

In the floss manual we have the example at the end of the mail and my questions are:

- how often are the offset and early variables updated? every k-cycle?
- is only one of these (offset and early) > 0 ?
- what is the `early` variable representing?
- are there multiple k-cycle running out of sync at the same time?

Because when i got it right, the mechanism is working like this:
- a main csound k-cycle is running
- when the accurate flag is given, a new instrument is not starting in sync with the main k-cycle, but to the nearest sample in time; so there are two k-cycles now running which are not in sync
- Csound gives the information at which point in the main-csound-k-cycle (?) the instrument is started and represents this in the offset and early variables
- so when the the new instrument is started at the 5th sample of the main-k-cycle the offset is 5
- but what is now happening with the early variable?

- Philipp

    int newopc_process_audio(CSOUND *csound, newopc *p){
      int i, n = CS_KSMPS;
      MYFLT *aout = p->out;  /* output signal */
      MYFLT cnt = p->var1 + *(p->in2);
      uint32_t offset = p->h.insdshead->ksmps_offset;
      uint32_t early  = p->h.insdshead->ksmps_no_end;

      /* sample-accurate mode mechanism */
      if(offset)
        memset(aout, '', offset*sizeof(MYFLT));
      if(early) {
        n -= early;
        memset(&aout[n], '', early*sizeof(MYFLT));
      }

      if(cnt > p->var2)
        cnt = (MYFLT) 0; /* check bounds */

      /* processing loop    */
      for(i=offset; i < n; i++)
        aout[i] = *(p->in1) + cnt;

      p->var1 = cnt; /* keep the value of cnt */
      return OK;
    }

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        Issues · csound/csound · GitHub
Discussions of bugs and features can be posted here

2

Offset may be nonzero at the start of the event and early may be nonzero at the end of the event.

There is boilerplate code in the opcode_sdk repository for dealing with this in the mult.c example.

Prof. Victor Lazzarini
Maynooth University
Ireland

3

ah, now i guess i got it:

offset is only important, when a event is starting.
so it creates a offset, just for the first k-cycle and only for this k-cycle it may be nonzero.

early is only important, when a event is stopped.
so it’s making the buffer writing process stop at the time, the instrument is stopped, and the remaining samples of the k-cycle are set to zero.

Thanks!