File: AudioFileReaderTest.h

package info (click to toggle)
sonic-visualiser 2.5~repack1-4
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 13,596 kB
  • ctags: 9,853
  • sloc: cpp: 93,843; ansic: 1,138; sh: 1,012; xml: 64; makefile: 35
file content (196 lines) | stat: -rw-r--r-- 6,430 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
/* -*- 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 2013 Chris Cannam.
    
    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.
*/

#ifndef TEST_AUDIO_FILE_READER_H
#define TEST_AUDIO_FILE_READER_H

#include "../AudioFileReaderFactory.h"
#include "../AudioFileReader.h"

#include "AudioTestData.h"

#include <cmath>

#include <QObject>
#include <QtTest>
#include <QDir>

#include <iostream>

using namespace std;

static QString audioDir = "testfiles";

class AudioFileReaderTest : public QObject
{
    Q_OBJECT

    const char *strOf(QString s) {
        return strdup(s.toLocal8Bit().data());
    }

private slots:
    void init()
    {
        if (!QDir(audioDir).exists()) {
            cerr << "ERROR: Audio test file directory \"" << audioDir << "\" does not exist" << endl;
            QVERIFY2(QDir(audioDir).exists(), "Audio test file directory not found");
        }
    }

    void read_data()
    {
        QTest::addColumn<QString>("audiofile");
        QStringList files = QDir(audioDir).entryList(QDir::Files);
        foreach (QString filename, files) {
            QTest::newRow(strOf(filename)) << filename;
        }
    }

    void read()
    {
        QFETCH(QString, audiofile);

        sv_samplerate_t readRate = 48000;

	AudioFileReader *reader =
	    AudioFileReaderFactory::createReader
	    (audioDir + "/" + audiofile, readRate);

        QStringList fileAndExt = audiofile.split(".");
        QStringList bits = fileAndExt[0].split("-");
        QString extension = fileAndExt[1];
        sv_samplerate_t nominalRate = bits[0].toInt();
        int nominalChannels = bits[1].toInt();
        int nominalDepth = 16;
        if (bits.length() > 2) nominalDepth = bits[2].toInt();

	if (!reader) {
#if ( QT_VERSION >= 0x050000 )
	    QSKIP("Unsupported file, skipping");
#else
	    QSKIP("Unsupported file, skipping", SkipSingle);
#endif
	}

        QCOMPARE((int)reader->getChannelCount(), nominalChannels);
        QCOMPARE(reader->getNativeRate(), nominalRate);
        QCOMPARE(reader->getSampleRate(), readRate);

	int channels = reader->getChannelCount();
	AudioTestData tdata(readRate, channels);
	
	float *reference = tdata.getInterleavedData();
        sv_frame_t refFrames = tdata.getFrameCount();
	
	// The reader should give us exactly the expected number of
	// frames, except for mp3/aac files. We ask for quite a lot
	// more, though, so we can (a) check that we only get the
	// expected number back (if this is not mp3/aac) or (b) take
	// into account silence at beginning and end (if it is).
	vector<float> test = reader->getInterleavedFrames(0, refFrames + 5000);
	sv_frame_t read = test.size() / channels;

        if (extension == "mp3" || extension == "aac" || extension == "m4a") {
            // mp3s and aacs can have silence at start and end
            QVERIFY(read >= refFrames);
        } else {
            QCOMPARE(read, refFrames);
        }

        // Our limits are pretty relaxed -- we're not testing decoder
        // or resampler quality here, just whether the results are
        // plainly wrong (e.g. at wrong samplerate or with an offset)

	double limit = 0.01;
        double edgeLimit = limit * 10; // in first or final edgeSize frames
        int edgeSize = 100; 

        if (nominalDepth < 16) {
            limit = 0.02;
        }
        if (extension == "ogg" || extension == "mp3" ||
            extension == "aac" || extension == "m4a") {
            limit = 0.2;
            edgeLimit = limit * 3;
        }

        // And we ignore completely the last few frames when upsampling
        int discard = 1 + int(round(readRate / nominalRate));

        int offset = 0;

        if (extension == "aac" || extension == "m4a") {
            // our m4a file appears to have a fixed offset of 1024 (at
            // file sample rate)
            offset = int(round((1024 / nominalRate) * readRate));
        }

        if (extension == "mp3") {
            // while mp3s appear to vary
            for (int i = 0; i < read; ++i) {
                bool any = false;
                double thresh = 0.01;
                for (int c = 0; c < channels; ++c) {
                    if (fabs(test[i * channels + c]) > thresh) {
                        any = true;
                        break;
                    }
                }
                if (any) {
                    offset = i;
                    break;
                }
            }
//            std::cerr << "offset = " << offset << std::endl;
        }

	for (int c = 0; c < channels; ++c) {
	    float maxdiff = 0.f;
	    int maxAt = 0;
	    float totdiff = 0.f;
	    for (int i = 0; i < read - offset - discard && i < refFrames; ++i) {
		float diff = fabsf(test[(i + offset) * channels + c] -
				   reference[i * channels + c]);
		totdiff += diff;
                // in edge areas, record this only if it exceeds edgeLimit
                if (i < edgeSize || i + edgeSize >= read - offset) {
                    if (diff > edgeLimit && diff > maxdiff) {
                        maxdiff = diff;
                        maxAt = i;
                    }
                } else {
                    if (diff > maxdiff) {
                        maxdiff = diff;
                        maxAt = i;
                    }
		}
	    }
	    float meandiff = totdiff / float(read);
//	    cerr << "meandiff on channel " << c << ": " << meandiff << endl;
//	    cerr << "maxdiff on channel " << c << ": " << maxdiff << " at " << maxAt << endl;
            if (meandiff >= limit) {
		cerr << "ERROR: for audiofile " << audiofile << ": mean diff = " << meandiff << " for channel " << c << endl;
                QVERIFY(meandiff < limit);
            }
	    if (maxdiff >= limit) {
		cerr << "ERROR: for audiofile " << audiofile << ": max diff = " << maxdiff << " at frame " << maxAt << " of " << read << " on channel " << c << " (mean diff = " << meandiff << ")" << endl;
		QVERIFY(maxdiff < limit);
	    }
	}
    }
};

#endif