Spectral Subtraction (SS) for speech enhancement

C implementation of spectral subtraction for speech enhancement

Table of contents

1. Development environments
2. Usages
3. Performance

Development environments

• Coding language: C language
• Integrated Development Environment(IDE): Microsoft Visual Studio 2017

Usages

• Noisy signal file name: speech_noisy.raw (Sampling rate: 8000HZ)
• Desirged output signal file name: speech_cleaned.raw
• Frame size: 256
• Base frame size to estimate noise signal: 5

User-dir:\SpeechEnhancement\Release> SpeechEnhancement.exe speech_noisy.raw speech_cleaned.raw 256 5

Performance

• Perceptual Evaluation of Speech Quality (PESQ): 3.35