view audio/ClipMixer.cpp @ 588:d122d3595a32

Store aggregate models in the document and release them when they are invalidated (because their components have been released). They're no longer leaked, but we still don't save them in the session file.

/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */

/*
    Sonic Visualiser
    An audio file viewer and annotation editor.
    Centre for Digital Music, Queen Mary, University of London.
    This file copyright 2006 Chris Cannam, 2006-2014 QMUL.
    
    This program is free software; you can redistribute it and/or
    modify it under the terms of the GNU General Public License as
    published by the Free Software Foundation; either version 2 of the
    License, or (at your option) any later version.  See the file
    COPYING included with this distribution for more information.
*/

#include "ClipMixer.h"

#include <sndfile.h>
#include <cmath>

#include "base/Debug.h"

//#define DEBUG_CLIP_MIXER 1

ClipMixer::ClipMixer(int channels, sv_samplerate_t sampleRate, sv_frame_t blockSize) :
    m_channels(channels),
    m_sampleRate(sampleRate),
    m_blockSize(blockSize),
    m_clipData(0),
    m_clipLength(0),
    m_clipF0(0),
    m_clipRate(0)
{
}

ClipMixer::~ClipMixer()
{
    if (m_clipData) free(m_clipData);
}

void
ClipMixer::setChannelCount(int channels)
{
    m_channels = channels;
}

bool
ClipMixer::loadClipData(QString path, double f0, double level)
{
    if (m_clipData) {
        cerr << "ClipMixer::loadClipData: Already have clip loaded" << endl;
        return false;
    }

    SF_INFO info;
    SNDFILE *file;
    float *tmpFrames;
    sv_frame_t i;

    info.format = 0;
    file = sf_open(path.toLocal8Bit().data(), SFM_READ, &info);
    if (!file) {
	cerr << "ClipMixer::loadClipData: Failed to open file path \""
             << path << "\": " << sf_strerror(file) << endl;
	return false;
    }

    tmpFrames = (float *)malloc(info.frames * info.channels * sizeof(float));
    if (!tmpFrames) {
        cerr << "ClipMixer::loadClipData: malloc(" << info.frames * info.channels * sizeof(float) << ") failed" << endl;
        return false;
    }

    sf_readf_float(file, tmpFrames, info.frames);
    sf_close(file);

    m_clipData = (float *)malloc(info.frames * sizeof(float));
    if (!m_clipData) {
        cerr << "ClipMixer::loadClipData: malloc(" << info.frames * sizeof(float) << ") failed" << endl;
	free(tmpFrames);
	return false;
    }

    for (i = 0; i < info.frames; ++i) {
	int j;
	m_clipData[i] = 0.0f;
	for (j = 0; j < info.channels; ++j) {
	    m_clipData[i] += tmpFrames[i * info.channels + j] * float(level);
	}
    }

    free(tmpFrames);

    m_clipLength = info.frames;
    m_clipF0 = f0;
    m_clipRate = info.samplerate;

    return true;
}

void
ClipMixer::reset()
{
    m_playing.clear();
}

double
ClipMixer::getResampleRatioFor(double frequency)
{
    if (!m_clipData || !m_clipRate) return 1.0;
    double pitchRatio = m_clipF0 / frequency;
    double resampleRatio = m_sampleRate / m_clipRate;
    return pitchRatio * resampleRatio;
}

sv_frame_t
ClipMixer::getResampledClipDuration(double frequency)
{
    return sv_frame_t(ceil(double(m_clipLength) * getResampleRatioFor(frequency)));
}

void
ClipMixer::mix(float **toBuffers, 
               float gain,
               std::vector<NoteStart> newNotes, 
               std::vector<NoteEnd> endingNotes)
{
    foreach (NoteStart note, newNotes) {
        if (note.frequency > 20 && 
            note.frequency < 5000) {
            m_playing.push_back(note);
        }
    }

    std::vector<NoteStart> remaining;

    float *levels = new float[m_channels];

#ifdef DEBUG_CLIP_MIXER
    cerr << "ClipMixer::mix: have " << m_playing.size() << " playing note(s)"
         << " and " << endingNotes.size() << " note(s) ending here"
         << endl;
#endif

    foreach (NoteStart note, m_playing) {

        for (int c = 0; c < m_channels; ++c) {
            levels[c] = note.level * gain;
        }
        if (note.pan != 0.0 && m_channels == 2) {
            levels[0] *= 1.0f - note.pan;
            levels[1] *= note.pan + 1.0f;
        }

        sv_frame_t start = note.frameOffset;
        sv_frame_t durationHere = m_blockSize;
        if (start > 0) durationHere = m_blockSize - start;

        bool ending = false;

        foreach (NoteEnd end, endingNotes) {
            if (end.frequency == note.frequency && 
                end.frameOffset >= start &&
                end.frameOffset <= m_blockSize) {
                ending = true;
                durationHere = end.frameOffset;
                if (start > 0) durationHere = end.frameOffset - start;
                break;
            }
        }

        sv_frame_t clipDuration = getResampledClipDuration(note.frequency);
        if (start + clipDuration > 0) {
            if (start < 0 && start + clipDuration < durationHere) {
                durationHere = start + clipDuration;
            }
            if (durationHere > 0) {
                mixNote(toBuffers,
                        levels,
                        note.frequency,
                        start < 0 ? -start : 0,
                        start > 0 ?  start : 0,
                        durationHere,
                        ending);
            }
        }

        if (!ending) {
            NoteStart adjusted = note;
            adjusted.frameOffset -= m_blockSize;
            remaining.push_back(adjusted);
        }
    }

    delete[] levels;

    m_playing = remaining;
}

void
ClipMixer::mixNote(float **toBuffers,
                   float *levels,
                   float frequency,
                   sv_frame_t sourceOffset,
                   sv_frame_t targetOffset,
                   sv_frame_t sampleCount,
                   bool isEnd)
{
    if (!m_clipData) return;

    double ratio = getResampleRatioFor(frequency);
    
    double releaseTime = 0.01;
    sv_frame_t releaseSampleCount = sv_frame_t(round(releaseTime * m_sampleRate));
    if (releaseSampleCount > sampleCount) {
        releaseSampleCount = sampleCount;
    }
    double releaseFraction = 1.0/double(releaseSampleCount);

    for (sv_frame_t i = 0; i < sampleCount; ++i) {

        sv_frame_t s = sourceOffset + i;

        double os = double(s) / ratio;
        sv_frame_t osi = sv_frame_t(floor(os));

        //!!! just linear interpolation for now (same as SV's sample
        //!!! player). a small sinc kernel would be better and
        //!!! probably "good enough"
        double value = 0.0;
        if (osi < m_clipLength) {
            value += m_clipData[osi];
        }
        if (osi + 1 < m_clipLength) {
            value += (m_clipData[osi + 1] - m_clipData[osi]) * (os - double(osi));
        }
         
        if (isEnd && i + releaseSampleCount > sampleCount) {
            value *= releaseFraction * double(sampleCount - i); // linear ramp for release
        }

        for (int c = 0; c < m_channels; ++c) {
            toBuffers[c][targetOffset + i] += float(levels[c] * value);
        }
    }
}