private static void CutAndCombine(string mp3, string output, string wav, List <Range> range)
{
string listFile = wav + ".list";
using (StreamWriter sw = new StreamWriter(listFile)) {
for (int i = 0; i < range.Count; i++)
{
var r = range[i];
var file = wav + i + ".mp3";
FFMpeg.Split(mp3, file, r.begin / 10.0, r.end / 10.0);
sw.WriteLine("file '{0}'", file);
}
}
FFMpeg.Concat(listFile, output);
}
//static int fft_len = 128;
static void Main(string[] args)
{
var mp3Dir = args[0];
var outDir = args[1];
var tmpDir = outDir + "\\tmp\\";
Directory.CreateDirectory(tmpDir);
List <string> fileList = new List <string>();
List <string> mp3FileList = new List <string>();
List <string> outFileList = new List <string>();
List <double[, ]> ffts = new List <double[, ]>();
List <List <int> > hashs = new List <List <int> >();
var ranges = new List <List <Range> >();
int size = 20;
LSH.init(size, MyFFT.len / 2);
foreach (var f in new DirectoryInfo(mp3Dir).GetFiles())
{
if (f.Extension.ToLower() != ".mp3")
{
continue;
}
Console.Write("{0}", f.Name);
var wavFile = tmpDir + f.Name + ".wav";
var fftFile = wavFile + ".fft";
TimeLogger log = new TimeLogger();
//DateTime last = DateTime.Now;
///
/// 1. 转wav
///
if (!File.Exists(wavFile) && !File.Exists(fftFile)) //wav只是为了生成fft，如果fft已经有了，就不用wav了
{
FFMpeg.Mp3toWav(f.FullName, wavFile);
}
fileList.Add(wavFile);
mp3FileList.Add(f.FullName);
outFileList.Add(outDir + "\\" + f.Name);
log.Log(Console.Out, "\t{0:F1}");
///
/// 2. fft
///
if (!File.Exists(fftFile))
{
var a = MyFFT.ProcessWavArr(wavFile);
WriteArrayToFile(a, fftFile);
ffts.Add(a);
File.Delete(wavFile); //生成fft之后就可以删除wav了
}
else
{
ffts.Add(ReadArrayFromFile(fftFile));
}
log.Log(Console.Out, "\t{0:F1}");
///
/// 3. hash
///
var hashFile = wavFile + ".hash";
List <int> hash = new List <int>();
if (!File.Exists(hashFile))
{
var a = ffts.Last();
using (StreamWriter sw = new StreamWriter(hashFile)) {
for (int i = 0; i + size < a.GetLength(0); i++)
{
int h = LSH.hash(a, i);
sw.WriteLine(h);
hash.Add(h);
}
}
}
else
{
using (StreamReader sr = new StreamReader(hashFile)) {
while (!sr.EndOfStream)
{
hash.Add(int.Parse(sr.ReadLine()));
}
}
}
hashs.Add(hash);
log.Log(Console.Out, "\t{0:F1}");
ranges.Add(new List <Range>());
for (int i = 0; i < fileList.Count - 1; i++)
{
int j = fileList.Count - 1;
var ranges_i = new List <Range>();
var ranges_j = new List <Range>();
var lst = CheckSame(ffts[i], ffts[j], hashs[i], hashs[j], size);
FineTune(ffts[i], ffts[j], hashs[i], hashs[j], ranges_i, ranges_j, size, lst);
//TODO 加上反向的看看效果会不会有变化
ranges_i = CompresssRange(ranges_i);
ranges_j = CompresssRange(ranges_j);
ranges[i] = CombineToRanges(ranges[i], ranges_i);
ranges[j] = CombineToRanges(ranges[j], ranges_j);
}
//for (int j = 0; j < fileList.Count; j++) {
// ranges[j] = CompresssRange(ranges[j]);
//}
//存储广告位置
for (int i = 0; i < fileList.Count; i++)
{
using (StreamWriter sw = new StreamWriter(fileList[i] + ".range")) {
//sw.WriteLine(ranges[i].Count);
foreach (var r in ranges[i])
{
sw.WriteLine("{0} {1} {2}", r.begin, r.end, r.count);
}
}
}
log.Log(Console.Out, "\t{0:F1}\n");
}
//读取广告位置，并且报告出现次数
for (int i = 0; i < fileList.Count; i++)
{
var range_file = fileList[i] + ".range";
using (StreamReader sr = new StreamReader(range_file)) {
var range = new List <Range>();
while (!sr.EndOfStream)
{
var x = sr.ReadLine().Split(' ');
if (x.Length != 3)
{
continue;
}
range.Add(new Range(int.Parse(x[0]), int.Parse(x[1]), int.Parse(x[2])));
}
ranges[i] = ReverseRange(range, ffts[i].GetLength(0));
}
}
//5. 提取、拼接 正文剩下部分
for (int i = 0; i < fileList.Count; i++)
{
CutAndCombine(mp3FileList[i], outFileList[i], fileList[i], ranges[i]);
}
//0.1s的精度。22.05khz
}
const bool saveTmpFile = false; //会不会输出临时文件。输出的话，第二次计算会快；不输出当然就省空间了
/// <summary>
/// 检测一系列MP3文件包含的广告（重复出现的部分）
/// </summary>
/// <param name="mp3Files"></param>
/// <param name="notify">在需要的时候调用notify()更新进度，参数是[0, 1]的浮点数</param>
/// <returns>对应各个MP3的广告区间</returns>
public static List <KeyValuePair <List <Range>, int> > DetectAD(string[] mp3Files, Action <int, int> notify)
{
var tmpDir = Path.Combine(Path.GetTempPath(), tmpdirName);
Directory.CreateDirectory(tmpDir);
List <string> fileList = new List <string>();
List <double[, ]> ffts = new List <double[, ]>();
List <List <int> > hashs = new List <List <int> >();
var ranges = new List <List <Range> >();
int size = 20;
LSH.init(size, MyFFT.len / 2);
List <Link> links = new List <Link>();
TimeEstimater log = new TimeEstimater();
log.InitType("wav", mp3Files.Length);
log.InitType("fft", mp3Files.Length);
log.InitType("hash", mp3Files.Length);
log.InitType("compare", mp3Files.Length * (mp3Files.Length - 1) / 2);
foreach (var fn in mp3Files)
{
var f = new FileInfo(fn);
if (f.Extension.ToLower() != ".mp3")
{
continue;
}
Console.Write("{0}", f.Name);
var wavFile = Path.Combine(tmpDir, f.Name + ".wav");
var fftFile = wavFile + ".fft";
log.StartTimer("wav");
///
/// 1. 转wav
///
if (!File.Exists(wavFile) && !File.Exists(fftFile)) //wav只是为了生成fft，如果fft已经有了，就不用wav了
{
FFMpeg.Mp3toWav(f.FullName, wavFile);
}
fileList.Add(wavFile);
log.EndTimer("wav");
notify((int)log.GetTimeUsed().TotalSeconds, (int)log.EstimateTime().TotalSeconds);
log.StartTimer("fft");
///
/// 2. fft
///
if (!File.Exists(fftFile))
{
var a = MyFFT.ProcessWavArr(wavFile);
if (saveTmpFile)
{
WriteArrayToFile(a, fftFile);
}
ffts.Add(a);
File.Delete(wavFile); //生成fft之后就可以删除wav了
}
else
{
ffts.Add(ReadArrayFromFile(fftFile));
}
log.EndTimer("fft");
notify((int)log.GetTimeUsed().TotalSeconds, (int)log.EstimateTime().TotalSeconds);
log.StartTimer("hash");
///
/// 3. hash
///
var hashFile = wavFile + ".hash";
List <int> hash = new List <int>();
if (!File.Exists(hashFile))
{
var a = ffts.Last();
for (int i = 0; i + size < a.GetLength(0); i++)
{
int h = LSH.hash(a, i);
hash.Add(h);
}
if (saveTmpFile)
{
using (StreamWriter sw = new StreamWriter(hashFile)) {
foreach (var h in hash)
{
sw.WriteLine(h);
}
}
}
}
else
{
using (StreamReader sr = new StreamReader(hashFile)) {
while (!sr.EndOfStream)
{
hash.Add(int.Parse(sr.ReadLine()));
}
}
}
hashs.Add(hash);
log.EndTimer("hash");
notify((int)log.GetTimeUsed().TotalSeconds, (int)log.EstimateTime().TotalSeconds);
//log.Log(Console.Out, "\t{0:F1}");
ranges.Add(new List <Range>());
for (int i = 0; i < fileList.Count - 1; i++)
{
int j = fileList.Count - 1;
log.StartTimer("compare");
var ranges_i = new List <Range>();
var ranges_j = new List <Range>();
var lst = CheckSame(ffts[i], ffts[j], hashs[i], hashs[j], size);
var tlink = FineTune(ffts[i], ffts[j], hashs[i], hashs[j], ranges_i, ranges_j, size, lst);
//TODO 加上反向的看看效果会不会有变化
//添加等价关系
foreach (var link in tlink)
{
links.Add(new Link(i, j, link.Key, link.Value));
}
ranges_i = CompresssRange(ranges_i);
ranges_j = CompresssRange(ranges_j);
ranges[i] = CombineToRanges(ranges[i], ranges_i);
ranges[j] = CombineToRanges(ranges[j], ranges_j);
log.EndTimer("compare");
notify((int)log.GetTimeUsed().TotalSeconds, (int)log.EstimateTime().TotalSeconds);
}
//存储广告位置
//for (int i = 0; i < fileList.Count; i++) {
// using (StreamWriter sw = new StreamWriter(fileList[i] + ".range")) {
// //sw.WriteLine(ranges[i].Count);
// foreach (var r in ranges[i]) {
// sw.WriteLine("{0} {1} {2}", r.begin, r.end, r.count);
// }
// }
//}
//log.Log(Console.Out, "\t{0:F1}\n");
}
for (int j = 0; j < ranges.Count; j++)
{
ranges[j] = CompresssRange(ranges[j]);
}
CalcRangeTypes(ranges, links);
//读取广告位置，并且报告出现次数
//for (int i = 0; i < fileList.Count; i++) {
// var range_file = fileList[i] + ".range";
// using (StreamReader sr = new StreamReader(range_file)) {
// var range = new List<Range>();
// while (!sr.EndOfStream) {
// var x = sr.ReadLine().Split(' ');
// if (x.Length != 3) continue;
// range.Add(new Range(int.Parse(x[0]), int.Parse(x[1]), int.Parse(x[2])));
// }
// ranges[i] = ReverseRange(range, ffts[i].GetLength(0));
// }
//}
var ret = new List <KeyValuePair <List <Range>, int> >();
for (int i = 0; i < ranges.Count; i++)
{
ret.Add(new KeyValuePair <List <Range>, int>(ranges[i], ffts[i].GetLength(0)));
}
return(ret);
}