SilenceRemovalConverter

Author: pschatzmann

examples/CMakeLists.txt

@@ -29,6 +29,7 @@ add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/time-to-speech")
 add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/unit-to-text")
 add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/number-to-speech")
 add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/unit-to-speech")
+add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/audio-i2s-nosilence")
 # not supported (yet?)
 #add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/time-to-speech-sd")
 #add_subdirectory( "${CMAKE_CURRENT_SOURCE_DIR}/talking-clock")

examples/audio-i2s-nosilence/CMakeLists.txt

@@ -0,0 +1,22 @@
+cmake_minimum_required(VERSION 3.20)
+
+# set the project name
+project(audio-i2s-nosilence)
+set (CMAKE_CXX_STANDARD 11)
+set (DCMAKE_CXX_FLAGS "-Werror")
+if (CMAKE_CXX_COMPILER_ID STREQUAL "Clang")
+    set (CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -fno-omit-frame-pointer -fsanitize=address")
+    set (CMAKE_LINKER_FLAGS_DEBUG "${CMAKE_LINKER_FLAGS_DEBUG} -fno-omit-frame-pointer -fsanitize=address")
+endif()
+
+include(FetchContent)
+
+# build sketch as executable
+add_executable (audio-i2s-nosilence audio-i2s-nosilence.ino ../main.cpp)
+# defines needed for desktop build
+target_compile_definitions(audio-i2s-nosilence PUBLIC -DARDUINO -DEXIT_ON_STOP -DIS_DESKTOP -DUSE_HELIX)
+# define ino as cpp
+set_source_files_properties(audio-i2s-nosilence.ino PROPERTIES LANGUAGE CXX)
+# specify libraries
+target_link_libraries(audio-i2s-nosilence arduino_emulator arduino_helix arduino-audio-tools simple-tts portaudio)
+

examples/audio-i2s-nosilence/Desktop.h

@@ -0,0 +1,11 @@
+#pragma once
+/**
+ * @brief Some special rules for desktop builds. 
+ * The output should go to PortAudioStream
+ */
+
+#ifdef IS_DESKTOP
+#include "AudioLibs/PortAudioStream.h"
+#define I2SStream PortAudioStream 
+#define AudioKitStream PortAudioStream
+#endif

examples/audio-i2s-nosilence/audio-i2s-nosilence.ino

@@ -11,11 +11,12 @@
 
 #include "SimpleTTS.h"
 #include "AudioCodecs/CodecMP3Helix.h"
+#include "Desktop.h"
 
-I2SStream i2s; // final output
-VolumeStream volume(i2s); // volume support
-SilenceRemovalConverter<int16_t> silence(8,2); // suppress silence
-ConvertedStream<int16_t, SilenceRemovalConverter<int16_t>> out(volume,silence);
+I2SStream i2s;
+VolumeStream volume(i2s);
+SilenceRemovalConverter<int16_t> rem(8, 2);
+ConvertedStream<int16_t,SilenceRemovalConverter<int16_t>> out(volume, rem); 
 
 MP3DecoderHelix mp3;
 AudioDictionary dictionary(ExampleAudioDictionaryValues);

src/SilenceRemovalStream.h

@@ -26,8 +26,9 @@ class SilenceRemovalStream : public audio_tools::AudioStreamX {
     LOGI("begin(n=%d, aplidudeLimit=%d", n, aplidudeLimit);
     this->n = n;
     this->amplidude_limit = aplidudeLimit;
-    this->priorLastAudioPos = 0;
+    this->priorLastAudioPos = n+1;  // ignore first values
     this->active = n > 0;
+    return true;
   }
 
   virtual size_t write(const uint8_t *data, size_t size) override {
@@ -53,18 +54,18 @@ class SilenceRemovalStream : public audio_tools::AudioStreamX {
     // write audio data w/o silence
     size_t write_size = write_count * sizeof(T);
     p_out->write(buffer, write_size);
-    LOGD("filtered from %d -> %d", size, write_size);
+    LOGI("filtered silence from %d -> %d", (int)size, (int)write_size);
 
-    // calculate negative offset for next run
-    priorLastAudioPos = size - findLastAudioPos(audio, sample_count - 1);
+    // number of empty samples of prior buffer
+    priorLastAudioPos =  findLastAudioPos(audio, sample_count - 1);
 
     // return processed data size
     return size;
   }
 
   virtual int availableForWrite() override { return p_out->availableForWrite(); }
 
-  void end() {
+  void end() override {
     priorLastAudioPos = 0;
     active = false;
   }
@@ -87,8 +88,8 @@ class SilenceRemovalStream : public audio_tools::AudioStreamX {
   int findLastAudioPos(T *audio, int pos) {
     for (int j = 0; j < n; j++) {
       // we are before the start of the current buffer
-      if (pos - j < 0) {
-        return pos - j + priorLastAudioPos;
+      if (pos - j <= 0) {
+        return priorLastAudioPos;
       }
       // we are in the current buffer
       if (abs(audio[pos - j]) > amplidude_limit) {