Re: [fluid-dev] fluid_sequencer_send_at ignoring time when not using an audio driver

[Pedro Lopez-Cabanillas]

On Sunday 17 June 2012, you wrote:
> On Sun 17 Jun 2012 13:02:39 IST, Pedro Lopez-Cabanillas wrote:
> > On Thursday 14 June 2012, Raja Mukherji wrote:
> >> I'm not sure if I'm doing something wrong but it seems as though
> >> fluid_sequencer_send_at is ignoring the time parameter.
> >> 
> >> I'm trying to use fluid_synth_write_float to render music to a buffer
> >> that I can then encode and stream to a browser. Hence realtime
> >> rendering is not important, rather rendering as fast as possible is
> >> desired.
> >> I've created a sequencer using new_fluid_sequencer2(0) so it's not
> >> using the system timer. I'm trying to render in 3 second blocks
> >> (arbitrary for now, I've tried smaller and longer blocks). For each
> >> block, I create and send all the events for that block. Then I call
> >> fluid_synth_write_float. It seems that the tick value of the sequencer
> >> does not change in a manner consistent with the call to
> >> fluid_synth_write_float so I am adjusting the times for all events in
> >> a block by the value returned by fluid_sequencer_get_tick and using
> >> relative times (so the last parameter to fluid_sequencer_send_at is
> >> false). I have tried using absolute times too, with no difference.
> > 
> > I can't reproduce your problem. Attached is a slightly version of our
> > fluidsynth_metronome program that is distributed in the doc/ folder of
> > FluidSynth:
> > http://fluidsynth.sourceforge.net/api/fluidsynth__metronome.c-example.h
> > tml
> > 
> > Instead of using an audio driver, this modified program renders 10 bars
> > of the metronome pattern to a wav disk file. It uses absolute tick
> > times, like the original sample program. The result has all beats
> > where they should be.
> > 
> > Regards,
> > Pedro
> 
> Thanks for your reply. I had originally used this this method with the
> sequencer callback and it worked fine, as it does in your example.

So, the problem is not in fluid_sequencer_send_at(), but in another place.

> The
> problem is that this method does suit the rest of the application which
> is processing events for any number of synths and for lv2 and other
> filters (basically a modular synth application). So I'm not using
> fluid_sequencer_register_client and a callback to add new events,
> instead I'm adding events using fluid_sequencer_send_at and then
> directly calling fluid_synth_write_float in effectively one large loop.

The length of the block processed by fluid_synth_write_float() is the same 
or smaller than the total duration of the events previously sent to the 
sequencer ?
 
> The code for the loop is spread over several files, but I've attached
> pseudo-code showing the effective loop.

Sorry, there are things that I don't understand in your pseudocode. For 
instance, how is calculated TIME_STEP in the call to 
fluid_synth_write_float() ? maybe you wanted to use FRAME_STEP instead ?

Can you please post a simple C or C++ working code that can be compiled and 
illustrates your problem?

Regards,
Pedro