better sound level detection algorithm

more options

1 files changed, 27 insertions, 26 deletions

diff --git a/sound_detector/main.cpp b/sound_detector/main.cpp
index 77557d5..9cb67e7 100644
--- a/sound_detector/main.cpp
+++ b/sound_detector/main.cpp
@@ -8,6 +8,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <complex.h>
 
 //c++
 #include <vector>
@@ -25,7 +26,8 @@
 
 
 
-unsigned int precapture = 3, postcapture = 7, min_length = 1, thresold_percent = 5;
+unsigned int precapture = 3, postcapture = 7, min_length = 1, thresold = 0, stop_thresold = 0;
+float thresold_percent = 5.0, stop_thresold_percent = 5.0;
 unsigned rate = 0;
 
 bool sound_detected = false, debug = false;
@@ -128,23 +130,21 @@ void handle_data()
 	vorbis_params vparams;
 	while(true)
 	{
-		boost::this_thread::sleep(boost::posix_time::seconds(1));
+		boost::this_thread::sleep(boost::posix_time::milliseconds(100));
 		if(!sound_detected)
 		{
-			if(debug)
-				printf("buffer contain %lu, required %d\n", buffer.size(), rate*precapture);
 			if(buffer.size() > rate * precapture)
 			{
-				int thresold = (int)(((float)INT16_MAX/100.0)*(float)thresold_percent), noise = 0;
+/*				if(debug)
+					printf("buffer contain %lu, required %d\n", buffer.size(), rate*precapture); */
+				unsigned long level = 0;
 				lock.lock();
 				for(size_t i = 0; i < buffer.size(); i++)
-				{
-					if((buffer[i] > thresold) || (buffer[i] < -thresold))
-						noise++;
-				}
+					level += abs(buffer[i]);
+				level /= buffer.size();
 				if(debug)
-					printf("noise detected %d, noise required %d\n", noise, min_length * rate);
-				if(noise > ((min_length * rate))) //need to do some research
+					printf("level detected %lu, level required %d\n", level, thresold);
+				if(level > thresold)
 				{
 					printf("write started\n");
 					boost::filesystem::path p(out_dir);
@@ -163,24 +163,20 @@ void handle_data()
 		}
 		else
 		{
-			if(debug)
-				printf("buffer contain %lu, required %d\n", buffer.size(), rate*postcapture);
 			if(buffer.size() > rate * postcapture)
 			{
-				int thresold = (int)(((float)INT16_MAX/100.0)*(float)thresold_percent), silence = 0;
+/*				if(debug)
+					printf("buffer contain %lu, required %d\n", buffer.size(), rate*postcapture); */
+				unsigned long level = 0;
 				lock.lock();
 				for(size_t i = 0; i < buffer.size(); i++)
-				{
-					if((buffer[i] < thresold) || (buffer[i] > -thresold) )
-						silence++;
-					else
-						silence = 0;
-				}
+					level += abs(buffer[i]);
+				level /= buffer.size();
 				encode_data(vparams);
 				if(debug)
-					printf("silence detected %d, silence required %d\n", silence, postcapture * rate);
+					printf("level detected %lu, level lower than %d required to stop\n", level, stop_thresold);
 				lock.unlock();
-				if(silence > (rate * postcapture))
+				if(level < stop_thresold)
 				{
 					if(debug)
 						printf("write stopped\n");
@@ -238,12 +234,12 @@ int main(int argc, char **argv)
 	int opt = -1, device = -1;
 	if(argc == 1)
 	{
-		printf("usage:\n%s -h -l -f -v -d <dev number> -o <path> -p <sec> -P <sec> -m <sec> -s <percents>\n\t-h\tthis help message\n\t-l\tdevice list\n\t-d\tdevice number from device list\n\t-o\toutput directory\n\t-f\tfork to bakground\n\t-v\tverbose\n\t-P\tpre capture seconds\n\t-p\tpost capture seconds\n\t-m\tminimum capture length\n\t-s\tsound level\n", argv[0]);
+		printf("usage:\n%s -h -l -f -v -d <dev number> -o <path> -p <sec> -P <sec> -m <sec> -s <float percents> -S <float percents>\n\t-h\tthis help message\n\t-l\tdevice list\n\t-d\tdevice number from device list\n\t-o\toutput directory\n\t-f\tfork to bakground\n\t-v\tverbose\n\t-P\tpre capture seconds\n\t-p\tpost capture seconds\n\t-m\tminimum capture length\n\t-s\tsound level in float value 5,0%% default\n\t-S minimum sound level to stop recording in float value 5,0%% is default\n", argv[0]);
 		return 1;
 	}
 	bool _fork = false;
 	//const unsigned int pecapture = 3, postcapture = 7, min_length = 1, thresold_percent = 5;
-	while((opt = getopt(argc, argv, "hlfvp:P:m:s:d:o:")) != -1)
+	while((opt = getopt(argc, argv, "hlfvp:P:m:s:S:d:o:")) != -1)
 	{
 		switch(opt)
 		{
@@ -269,7 +265,10 @@ int main(int argc, char **argv)
 			min_length = atoi(optarg);
 			break;
 			case 's':
-			thresold_percent = atoi(optarg);
+			thresold_percent = strtof(optarg, NULL);
+			break;
+			case 'S':
+			stop_thresold_percent = strtof(optarg, NULL);
 			break;
 			case 'o':
 			strcpy(out_dir, optarg);
@@ -282,7 +281,7 @@ int main(int argc, char **argv)
 			break;
 			case 'h':
 			default:
-			printf("usage:\n%s -h -l -f -v -d <dev number> -o <path> -p <sec> -P <sec> -m <sec> -s <percents>\n\t-h\tthis help message\n\t-l\tdevice list\n\t-d\tdevice number from device list\n\t-o\toutput directory\n\t-f\tfork to bakground\n\t-v\tverbose\n\t-P\tpre capture seconds\n\t-p\tpost capture seconds\n\t-m\tminimum capture length\n\t-s\tsound level\n", argv[0]);
+			printf("usage:\n%s -h -l -f -v -d <dev number> -o <path> -p <sec> -P <sec> -m <sec> -s <float percents> -S <float percents>\n\t-h\tthis help message\n\t-l\tdevice list\n\t-d\tdevice number from device list\n\t-o\toutput directory\n\t-f\tfork to bakground\n\t-v\tverbose\n\t-P\tpre capture seconds\n\t-p\tpost capture seconds\n\t-m\tminimum capture length\n\t-s\tsound level in float value 5,0%% default\n\t-S minimum sound level to stop recording in float value 5,0%% is default\n", argv[0]);
 			break;
 		};
 	}
@@ -306,6 +305,8 @@ int main(int argc, char **argv)
 		printf("ERROR: output directory is not directory %%) \n");
 		return 1;
 	}
+	thresold = (int)(((float)INT16_MAX/100.0)*thresold_percent);
+	stop_thresold = (int)(((float)INT16_MAX/100.0)*stop_thresold_percent);
 	PaStream *stream;
 	PaStreamParameters params;
 	memset(&params, 0, sizeof(PaStreamParameters));