Florian, Just gave your dry and wet sample a quick listen. It sounds promising. I'm looking forward to giving it a more serious run. ron --- Florian Schmidt <mista.tapas@xxxxxxx> wrote: > > Hi, > > i implemented a small convolution engine for jack.. > grab it at > > http://affenbande.org/~tapas/jack_convolve-0.0.1.tgz > > untar > make > ./jack_convolve responsefile.wav > > It creates as many in/out ports as the response file > has channels. Uses > fftw3f and libsndfile. Will add libsamplerate > support in near future. > So for now, make sure the samplerate of jack and the > response file > match. > > Here's a ca 1 sec 48khz (resampled from 96khz) > stereo response file of a > room: > > http://affenbande.org/~tapas/FrontFacing%20TLM170.wav > > It's from this package: > > http://www.noisevault.com/index.php?page=3&action=file&file_id=130 > > which has 96khz responses.. > > Consumes ca. 25-30% cpu load on my 1.2ghz athlon at > jack buffer size 2048 > [;)] > > So there's plenty room for optimization (and some > return value checking > will be added too ;)).. If you know some tricks, let > me know.. The > sourcecode is pasted below for easier reference. > > Flo > > P.S.: thanks to mario lang for collaborating and > giving some hints > towards using fftw3f instead of fftw and some other > optimizations.. > > P.P.S.: oh yeah, example sound, here you go > [hydrogen dry then with > output of jack_convolve mixed to it]: > > http://affenbande.org/~tapas/jack_conv_ex1.ogg > > And here the convoluted signal alone: > > http://affenbande.org/~tapas/jack_conv_ex2.ogg > > P.P.S.: known issues: > > - won't handle samplerate or buffersize changes > gracefully > > - will bring your machine to a crawl ;) > > > jack_convolve.cc: > --------------- > > /* > Copyright (C) 2004 Florian Schmidt > > This program is free software; you can > redistribute it and/or modify > it under the terms of the GNU Lesser General > Public License as published by > the Free Software Foundation; either version 2.1 > of the License, or > (at your option) any later version. > > This program is distributed in the hope that it > will be useful, > but WITHOUT ANY WARRANTY; without even the > implied warranty of > MERCHANTABILITY or FITNESS FOR A PARTICULAR > PURPOSE. See the > GNU Lesser General Public License for more > details. > > You should have received a copy of the GNU > Lesser General Public License > along with this program; if not, write to the > Free Software > Foundation, Inc., 59 Temple Place - Suite 330, > Boston, MA 02111-1307, USA. > > $Id: types_8h-source.html,v 1.1 2004/04/27 > 18:21:48 joq Exp $ > */ > > > > #include <jack/jack.h> > #include <iostream> > #include <sstream> > #include <unistd.h> > #include <signal.h> > #include <stdio.h> > #include <sndfile.h> > #include <vector> > #include <cmath> > #include <fftw3.h> > > jack_client_t *client; > > std::vector<jack_port_t *> iports; > std::vector<jack_port_t *> oports; > > jack_nframes_t jack_buffer_size; > int chunks_per_channel; > > // channel chunk data > std::vector<std::vector <fftwf_complex*> > chunks; > > // the buffers for the fft > float *fft_float; > fftwf_complex *fft_complex; > > // the plan > fftwf_plan fft_plan_forward; > fftwf_plan fft_plan_backward; > > float normalize_factor; > > // per channel we need a ringbuffer holding the fft > results of the > // audio periods passed to us by jackd. > // each needs to be sized jack_buffer_size * > chunks_per_channel (see main()) > std::vector<fftwf_complex *> ringbuffers; > > // this gets advanced by jack_buffer_size after each > process() callback > unsigned int ringbuffer_index = 0; > > // a vector to hold the jack buffer pointers.. these > get resized > // during init in main > std::vector<jack_default_audio_sample_t *> ibuffers; > std::vector<jack_default_audio_sample_t *> obuffers; > > // channel data > std::vector<jack_default_audio_sample_t *>overlaps; > > int process(jack_nframes_t frames, void *arg) { > // std::cout << " " << ringbuffer_index; > // get pointer[s] to the buffer[s] > int channels = chunks.size(); > > for (int channel = 0; channel < channels; > ++channel) { > ibuffers[channel] = > ((jack_default_audio_sample_t*)jack_port_get_buffer(iports[channel], > frames)); > obuffers[channel] = > ((jack_default_audio_sample_t*)jack_port_get_buffer(oports[channel], > frames)); > } > > for (int channel = 0; channel < channels; > ++channel) { > // copy input buffer to fft buffer > for (int frame = 0; frame < jack_buffer_size; > ++frame) { > fft_float[frame] = > (float)(ibuffers[channel][frame]); > fft_float[frame+jack_buffer_size] = 0.0; > } > // fft the input[s] > fftwf_execute(fft_plan_forward); > > // store the new result into the ringbuffer for > this channel > for (int frame = 0; frame < jack_buffer_size * 2; > ++frame) { > ringbuffers[channel][ringbuffer_index+frame][0] = > fft_complex[frame][0] / normalize_factor; > ringbuffers[channel][ringbuffer_index+frame][1] = > fft_complex[frame][1] / normalize_factor; > } > > // zero our buffer for the inverse FFT, so we can > simply += the > // values in the next step. > for (int frame = 0; frame < jack_buffer_size * 2; > ++frame) { > fft_complex[frame][0] = 0; > fft_complex[frame][1] = 0; > } > > === message truncated === __________________________________ Do you Yahoo!? Take Yahoo! Mail with you! Get it on your mobile phone. http://mobile.yahoo.com/maildemo
