First, thanks to Steven and Rory's help, I've gotten tantalizingly close. So thanks! I can hear my score playing. =) 

The bad news is that there's a constant extra sound, sounds like a distorted wave that is quite loud, and always present, and I can hear the score quieter, but also badly distorted. Maybe it has something to do with buffer sizes or sample types? If anyone can tell from looking at this where I  might be wrong, that would be very helpful.

jack is set to a buffer size of 128 ( this runs no problem when not outputing through the spout )

argc/argv are:  -+rtaudio=null -b128 myorc.orc myscore.sco

Here's what I have in my engine class:

// setup csound
void Engine::initCsound(int argc, char **argv){
   cout << "Engine::initCsound()\n";
   csound = new Csound();
   // TODO: some error handling here, yo
   int result = csound->PreCompile();
   // tell csound not to handle audio i/o itself
   // the 128 matches the jack buff size and the csound buffsize 
   csound->SetHostImplementedAudioIO(1, 128);
   result = csound->Compile(argc, argv);
   // output needs to be scaled
   csoundScale = csound->Get0dBFS();
   // save ref to the csound spout buffer, csoundOuput is a MYFLT*
   csoundOutput = csound->GetSpout();
}
// perform a csound Buffer
MYFLT* Engine::performBuffer(){
   int result = csound->PerformBuffer();
   return csoundOutput;
}

And here's how it gets used in the jack audio process callback, (the csound orc is mono for simplicity)

int process (jack_nframes_t nframes, void *data){
   // cast the void pointer to our engine
   Engine *engine = (Engine *)data;
   // grab our output buffer 
   sample_t *out = (sample_t *) jack_port_get_buffer(output_port, nframes);
   // perform one csound buffer, this copies audio to engine->csoundOuput
   MYFLT *csoundOutput = engine->performBuffer();
   // iterate through the jack buffer copying output from csound
   for(jack_nframes_t i=0; i < nframes; i++){
      out[i] = (sample_t)( csoundOutput[i] * engine->csoundScale );
   }
   // let jack know all is a-ok
   return 0;  
}

Any advice would be wonderful!

Iain

I am just reading your code and I see you have this:

csoundOutput = csound->GetSpout();

which should be

csoundOutput = csound->GetOutputBuffer();

===
Otherwise

// perform a csound Buffer
MYFLT* Engine::performBuffer(){
  int result = csound->PerformBuffer();
  return csoundOutput;
}

needs to be

MYFLT* Engine::performBuffer(){
  int result = csound->PerformKsmps();
  return csoundOutput;
}

===

It's useful to keep 'result' though, as it will be non-zero if you reached the end of performance.

On 22 Dec 2011, at 07:09, Iain Duncan wrote:

> First, thanks to Steven and Rory's help, I've gotten tantalizingly close. So thanks! I can hear my score playing. =)
>
> The bad news is that there's a constant extra sound, sounds like a distorted wave that is quite loud, and always present, and I can hear the score quieter, but also badly distorted. Maybe it has something to do with buffer sizes or sample types? If anyone can tell from looking at this where I  might be wrong, that would be very helpful.
>
> jack is set to a buffer size of 128 ( this runs no problem when not outputing through the spout )
> argc/argv are:  -+rtaudio=null -b128 myorc.orc myscore.sco
>
> Here's what I have in my engine class:
> // setup csound
> void Engine::initCsound(int argc, char **argv){
>    cout << "Engine::initCsound()\n";
>    csound = new Csound();
>    // TODO: some error handling here, yo
>    int result = csound->PreCompile();
>    // tell csound not to handle audio i/o itself
>    // the 128 matches the jack buff size and the csound buffsize
>    csound->SetHostImplementedAudioIO(1, 128);
>    result = csound->Compile(argc, argv);
>    // output needs to be scaled
>    csoundScale = csound->Get0dBFS();
>    // save ref to the csound spout buffer, csoundOuput is a MYFLT*
>    csoundOutput = csound->GetSpout();
> }
> // perform a csound Buffer
> MYFLT* Engine::performBuffer(){
>    int result = csound->PerformBuffer();
>    return csoundOutput;
> }
>
> And here's how it gets used in the jack audio process callback, (the csound orc is mono for simplicity)
>
> int process (jack_nframes_t nframes, void *data){
>    // cast the void pointer to our engine
>    Engine *engine = (Engine *)data;
>    // grab our output buffer
>    sample_t *out = (sample_t *) jack_port_get_buffer(output_port, nframes);
>    // perform one csound buffer, this copies audio to engine->csoundOuput
>    MYFLT *csoundOutput = engine->performBuffer();
>    // iterate through the jack buffer copying output from csound
>    for(jack_nframes_t i=0; i < nframes; i++){
>       out[i] = (sample_t)( csoundOutput[i] * engine->csoundScale );
>    }
>    // let jack know all is a-ok
>    return 0;
> }
>
>
> Any advice would be wonderful!
> Iain

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie

Send bugs reports to the Sourceforge bug tracker
           https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Thanks Victor, maybe when I get this figured out I could add to the comments in csound.h, I got confused reading them. Could you tell me if I'm correct that:

- spout, returned by csoundGetSpout,  gets filled with ( ksmp * number of channels ) samples after a call to csoundPerformKsmp. Or is it (ksmp * number of channels * -b) samples?

- the output buffer (not the same thing as spout), returned by csoundGetOutputBuffer, gets filled with ( -b * number of channels ) samples after a call to csoundPerformBuffer

thanks so much, will try them both out.

Iain

On Thu, Dec 22, 2011 at 12:04 PM, Victor Lazzarini <Victor.Lazzarini@nuim.ie> wrote:

I am just reading your code and I see you have this:

csoundOutput = csound->GetSpout();

which should be

csoundOutput = csound->GetOutputBuffer();

===
Otherwise

// perform a csound Buffer
MYFLT* Engine::performBuffer(){
  int result = csound->PerformBuffer();
  return csoundOutput;
}

needs to be

MYFLT* Engine::performBuffer(){
  int result = csound->PerformKsmps();
  return csoundOutput;
}

===

It's useful to keep 'result' though, as it will be non-zero if you reached the end of performance.

On 22 Dec 2011, at 07:09, Iain Duncan wrote:

> First, thanks to Steven and Rory's help, I've gotten tantalizingly close. So thanks! I can hear my score playing. =)
>
> The bad news is that there's a constant extra sound, sounds like a distorted wave that is quite loud, and always present, and I can hear the score quieter, but also badly distorted. Maybe it has something to do with buffer sizes or sample types? If anyone can tell from looking at this where I  might be wrong, that would be very helpful.
>
> jack is set to a buffer size of 128 ( this runs no problem when not outputing through the spout )
> argc/argv are:  -+rtaudio=null -b128 myorc.orc myscore.sco
>
> Here's what I have in my engine class:
> // setup csound
> void Engine::initCsound(int argc, char **argv){
>    cout << "Engine::initCsound()\n";
>    csound = new Csound();
>    // TODO: some error handling here, yo
>    int result = csound->PreCompile();
>    // tell csound not to handle audio i/o itself
>    // the 128 matches the jack buff size and the csound buffsize
>    csound->SetHostImplementedAudioIO(1, 128);
>    result = csound->Compile(argc, argv);
>    // output needs to be scaled
>    csoundScale = csound->Get0dBFS();
>    // save ref to the csound spout buffer, csoundOuput is a MYFLT*
>    csoundOutput = csound->GetSpout();
> }
> // perform a csound Buffer
> MYFLT* Engine::performBuffer(){
>    int result = csound->PerformBuffer();
>    return csoundOutput;
> }
>
> And here's how it gets used in the jack audio process callback, (the csound orc is mono for simplicity)
>
> int process (jack_nframes_t nframes, void *data){
>    // cast the void pointer to our engine
>    Engine *engine = (Engine *)data;
>    // grab our output buffer
>    sample_t *out = (sample_t *) jack_port_get_buffer(output_port, nframes);
>    // perform one csound buffer, this copies audio to engine->csoundOuput
>    MYFLT *csoundOutput = engine->performBuffer();
>    // iterate through the jack buffer copying output from csound
>    for(jack_nframes_t i=0; i < nframes; i++){
>       out[i] = (sample_t)( csoundOutput[i] * engine->csoundScale );
>    }
>    // let jack know all is a-ok
>    return 0;
> }
>
>
> Any advice would be wonderful!
> Iain

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie

Send bugs reports to the Sourceforge bug tracker
           https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

In PCM audio ("digital audio") of the sort that soundfiles and Csound
deal in, a sample is one scalar number. A sample frame is one sample
for every channel of audio. So a mono frame has 1 sample, a stereo
frame has 2 samples, a quad frame as 4 samples, and so on. When people
say "sampling rate" what they actually mean is "audio sample frames
per second" not "samples per second".

Buffers can be sized in raw bytes, samples, or frames. In addition,
there is a hierarchy of buffers of different sizes. The lowest level
buffer is the one that the synthesizer actually accumulates audio in
before sending that audio out. In Csound, this is the spout buffer.
Then, Csound has an intermediate buffer, and the audio output device
driver has its own buffer; the Csound intermediate buffer is some
integral fraction of the audio output device driver buffer.

Any or all of these buffers may be sized in raw bytes, samples, or
frames. E.g. spout for 2 channels at 100 ksmps for floating point
samples would be 100 frames (that is what ksmps are, frames), 200
samples, 800 bytes; with double-precision samples, that would be 100
frames, 200 samples, 1600 bytes.

Csound works with Jack both through my Jacko opcodes and through the
command-line jack option.

For the Jacko opcodes, the following conditions must obtain, these may
or may not shed light on the command-line Jack driver:

Csound's sr (actually, frames per second) must be equal to the Jack
daemon's frames per second.

Csound's ksmps (actually, frames per kperiod) must be equal to the
Jack daemon's frames per period.

Frames per period must not only (a) be a power of 2, but also (b) go
evenly into the frames per second, e.g. 128 frames per period goes
into 48000 frames per second 375 times, for a latency or MIDI time
granularity of about 2.7 milliseconds (as good as or better than the
absolute best human performers).

Jack is not as flexible as Csound when it comes to frames per period
etc., so first get Jack working to your satisfaction without Csound,
then configure Csound to work with Jack's frames per second and frames
per period.

Hope this helps,
Mike

On Thu, Dec 22, 2011 at 2:54 PM, Victor Lazzarini
<Victor.Lazzarini@nuim.ie> wrote:
> If you are using performBuffer(), you should access the output buffer
> (csoundGetOutputBuffer()); if you are using performKsmps(), you should
> access spout.
> The sizes are as follows: spout -> ksmps*nchnls samples
> OutputBuffer: csoundGetOutputBufferSize() samples
>
> Victor
>
> On 22 Dec 2011, at 19:33, Iain Duncan wrote:
>
> On Thu, Dec 22, 2011 at 11:24 AM, Victor Lazzarini
> <Victor.Lazzarini@nuim.ie> wrote:
>>
>> No, what I mean is that csound buffer sizes are counted in samples, not
>> frames. So, if you have a stereo buffer, size 128, it will have 64 frames.
>> Confusingly, spout/spin will be ksmps frames long. But your code is
>> reading from the buffer, so just watch out.
>> This is something that sometimes gets overlooked.
>
>
> Um, I'm feeling dense but I just got totally lost reading the above. I have
> (for now) only one channel of output in both jack and csound just to get
> things working first. When I use the stk in my host, I copy the output of
> one stk tick ( 1 sample ) into one jack frame and all sounds ok. With
> csound, I thought that calling csound->performBuffer meant that spout was
> now filled with one csound buffers worth of samples, 128 samples if -b128.
>
> Are you saying that after calling csound->PerformBuffer, I should read ksmp
> * -b samples from spout? is -b a buffer of ksmps? Or if not, can anyone tell
> me how much of what I need to copy? I'm not attached to using performBuffer,
> I could use performKsmp too if that would be easier. How many samples are in
> the spout after calling peformKsmp?
>
> thanks!
> Iain
>
>
> Dr Victor Lazzarini
> Senior Lecturer
> Dept. of Music
> NUI Maynooth Ireland
> tel.: +353 1 708 3545
> Victor dot Lazzarini AT nuim dot ie
>
>
>

--
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com

Send bugs reports to the Sourceforge bug tracker
           https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"