Fix audio normalizationHEADmaster

Normalization was putting audio onto range [0, 255], while it should
have been on range [0, 1].

* Add AudioBuffer::save() to enable debugging audio issues.

2 files changed, 48 insertions, 1 deletions

diff --git a/Whisper/MF/AudioBuffer.h b/Whisper/MF/AudioBuffer.h
index 11b5ead..6b5c264 100644
--- a/Whisper/MF/AudioBuffer.h
+++ b/Whisper/MF/AudioBuffer.h
@@ -1,5 +1,6 @@
 #pragma once
 #include <algorithm>
+#include <fstream>
 #include <vector>
 
 namespace Whisper
@@ -76,8 +77,40 @@ namespace Whisper
 			for (auto& elm : mono) {
 				elm -= min;
 				elm /= (max - min) + 1;
-				elm *= 255.0;
 			}
 		}
+
+		void save(const char* path, const int sample_rate) {
+			const int n_samples = mono.size();
+			const int bits_per_sample = sizeof(mono[0]) * 8;
+			const int n_channels = 1;
+			const int byte_rate = sample_rate * n_channels * bits_per_sample / 8;
+			const int block_align = n_channels * bits_per_sample / 8;
+			const int data_chunk_size = n_samples * n_channels * bits_per_sample / 8;
+			const int file_size = 36 + data_chunk_size;
+
+			std::ofstream ofs(path, std::ios::out | std::ios::binary);
+			ofs.write("RIFF", 4);
+			ofs.write((char*)&file_size, 4);
+			ofs.write("WAVE", 4);
+
+			ofs.write("fmt ", 4);
+			const int fmt_chunk_size = 16;
+			ofs.write((char*)&fmt_chunk_size, 4);
+			const short audio_format = 1;  // PCM
+			ofs.write((char*)&audio_format, 2);
+			ofs.write((char*)&n_channels, 2);
+			ofs.write((char*)&sample_rate, 4);
+			ofs.write((char*)&byte_rate, 4);
+			ofs.write((char*)&block_align, 2);
+			ofs.write((char*)&bits_per_sample, 2);
+
+			ofs.write("data", 4);
+			ofs.write((char*)&data_chunk_size, 4);
+			for (int i = 0; i < n_samples; i++) {
+				short sample = (short)(mono[i] * 32767.0f);
+				ofs.write((char*)&sample, 2);
+			}
+		};
 	};
 }
\ No newline at end of file
diff --git a/Whisper/Whisper/ContextImpl.capture.cpp b/Whisper/Whisper/ContextImpl.capture.cpp
index ce51393..2f29934 100644
--- a/Whisper/Whisper/ContextImpl.capture.cpp
+++ b/Whisper/Whisper/ContextImpl.capture.cpp
@@ -142,7 +142,21 @@ namespace
 			workStatus = S_FALSE;
 			buffer.currentOffset = pcmStartTime;
 			buffer.pcm = pcm;
+#if 0
+			{
+				static int i = 0;
+				std::string filename = "buf_" + std::to_string(i++) + "_raw.wav";
+				buffer.pcm.save(filename.c_str(), SAMPLE_RATE / 2);
+			}
+#endif
 			buffer.pcm.normalize();
+#if 0
+			{
+				static int i = 0;
+				std::string filename = "buf_" + std::to_string(i++) + "_normalized.wav";
+				buffer.pcm.save(filename.c_str(), SAMPLE_RATE / 2);
+			}
+#endif
 			SubmitThreadpoolWork( work );
 			pcmStartTime = nextSampleTime;
 			pcm.retainLast(captureParams.retainDuration);