[DEV] better display

analyser/appl/main.cpp

@@ -13,9 +13,37 @@
 #include <etk/theme/theme.hpp>
 #include <ejson/ejson.hpp>
 #include <audio/algo/speex/Resampler.hpp>
+#include <audio/algo/speex/Vad.hpp>
 
 static int resampling_quality = 10; // speex resampler quality
 
+static void usage() {
+	APPL_PRINT("  --corpus_root_path=XXX    Root path of the corpus (add after a folder with the name and other informations");
+	APPL_PRINT("  --corpus_output_path=XXX  DESTINATION of the data");
+	exit(0);
+}
+/*
+class InputData {
+	public:
+		etk::Vector<int16_t> stream;
+		ejson::Doc property;
+};
+
+InputData readInput(const etk::Uri& _uri) {
+	InputData out;
+	out.property.load(_uri);
+	
+	etk::String fileName = out.property["audio_filename"].toString().get();
+	etk::Uri fileNameAudioFile = it;
+	fileNameAudioFile.setPath(fileNameAudioFile.getPath().getParent() / fileName);
+	return out;
+}
+
+ejson::Doc readConfig(const etk::Uri& _uri) {
+	
+}
+*/
+
 /**
  * @brief Main of the program (This can be set in every case, but it is not used in Andoid...).
  * @param std IO
@@ -23,9 +51,28 @@ static int resampling_quality = 10; // speex resampler quality
  */
 int main(int _argc, const char *_argv[]) {
 	etk::init(_argc, _argv);
+	etk::Vector<etk::String> cmdLine;
+	for(int32_t iii=1 ; iii<_argc; iii++) {
+		cmdLine.pushBack(_argv[iii]);
+	}
+	etk::String corpusPath;
+	etk::String corpusOutput;
+	for(auto &it: cmdLine) {
+		etk::String tmpppp = it;
+		if (    tmpppp == "-h"
+		     || tmpppp == "--help") {
+			usage();
+		} else if (tmpppp.startWith("--corpus_root_path=") == true) {
+			corpusPath = tmpppp.extract(19);;
+		} else if (tmpppp.startWith("--corpus_output_path=") == true) {
+			corpusOutput = tmpppp.extract(21);;
+		} else {
+			APPL_ERROR("Unknow parameter: '" << tmpppp << "'" );
+			//usage();
+		}
+	}
 	
-	
+	etk::Vector<etk::Uri> list = etk::uri::listRecursive("FILE:///framework/musicdsp/audio-reco-corpus-french/corpus_2019_DupinEdouard_1983/");
-	etk::Vector<etk::Uri> list = etk::uri::listRecursive("FILE:///framework/musicdsp/audio-reco-corpus/corpus1/");
 	auto it = list.begin();
 	while (it != list.end()) {
 		if (it->getPath().getExtention() == "json") {
@@ -38,11 +85,16 @@ int main(int _argc, const char *_argv[]) {
 		APPL_DEBUG("elem=" << it);
 		ejson::Document doc;
 		doc.load(it);
+		/////////////////////////////////////////////////////
+		/// LOAD and convert in 16k if needed:
+		/////////////////////////////////////////////////////
 		etk::String fileName = doc["audio_filename"].toString().get();
 		etk::Uri fileNameAudioFile = it;
 		fileNameAudioFile.setPath(fileNameAudioFile.getPath().getParent() / fileName);
 		etk::Uri fileNameAudioFile16k = it;
 		fileNameAudioFile16k.setPath(fileNameAudioFile16k.getPath().getParent() / fileName + "_16k");
+		etk::Uri fileNameAudioFileVAD = it;
+		fileNameAudioFileVAD.setPath(fileNameAudioFileVAD.getPath().getParent() / fileName + "_VAD");
 		APPL_DEBUG("    file=" << fileName);
 		int32_t sampleRate = doc["audio_sample_rate"].toNumber().get();
 		APPL_DEBUG("    sampleRate=" << sampleRate);
@@ -85,7 +137,7 @@ int main(int _argc, const char *_argv[]) {
 				}
 			} else {
 				APPL_DEBUG("        *** LOAD_FILE ***");
-				// read stream: (We already have resample)
+				// read stream: (We already have resampled)
 				{
 					ememory::SharedPtr<etk::io::Interface> fileIO = etk::uri::get(fileNameAudioFile16k);
 					if (fileIO->open(etk::io::OpenMode::Read) == false) {
@@ -104,6 +156,56 @@ int main(int _argc, const char *_argv[]) {
 			dataBuffer = fileIO->readAll<int16_t>();
 			fileIO->close();
 		}
+		
+		//marque loom
+		
+		/////////////////////////////////////////////////////
+		/// Generate the VAD output:
+		/////////////////////////////////////////////////////
+		etk::Vector<int16_t> vadBuffer;
+		{
+			audio::algo::speex::Vad algo;
+			algo.init(1, 16000, audio::format_int16);
+			int16_t out[1024];
+			int16_t feedback[1024];
+			memset(feedback, 0, 2048);
+			for (int32_t iii=0; iii<dataBuffer.size()/32; ++iii) {
+				algo.process(out, &dataBuffer[iii*32], feedback, 32);
+				for (int32_t jjj=0; jjj<32; ++jjj) {
+					vadBuffer.pushBack(algo.getVoiceDetected()?32000:0);
+				}
+			}
+		}
+		APPL_DEBUG("        *** Store VAD ***  [BEGIN]");
+		{
+			ememory::SharedPtr<etk::io::Interface> fileIO = etk::uri::get(fileNameAudioFileVAD);
+			if (fileIO->open(etk::io::OpenMode::Write) == false) {
+				return -1;
+			}
+			fileIO->writeAll(vadBuffer);
+			fileIO->close();
+		}
+		APPL_DEBUG("        *** Store VAD ***  [ END ]");
+		
+		/////////////////////////////////////////////////////
+		/// Remove unneeded data (clip all unneeded data 
+		/// before and after the usable signal):
+		/////////////////////////////////////////////////////
+		
+		
+		
+		/////////////////////////////////////////////////////
+		/// Generate the FFT:
+		/////////////////////////////////////////////////////
+		
+		
+		
+		/////////////////////////////////////////////////////
+		/// Generate the CEPSTRE:
+		/////////////////////////////////////////////////////
+		
+		
+		
 	}
 	
 	return 0;

research.md

@@ -15,7 +15,7 @@ Natural Language Processing (NLP): https://github.com/zalandoresearch/flair
 ideas of architeture:
 =====================
 
-![objectuve work](work.svg)
+![objectuve work](work.jpg)