Scholarlink

<p>In wireless communication systems, the presence of noise significantly degrades the quality of transmitted speech signals. Our primary objective is to improve both the speech quality and intelligibility of coding systems in real-world noisy environments. This work presents an advanced approach for enhancing the robustness of linear predictive coding (LPC)-based speech coding system for various noise types and signal-to-noise ratio (SNR) levels. We begin by conducting a comprehensive evaluation of six noise suppression techniques to determine their efficacy using speech quality and intelligibility metrics. Following the evaluation, the <i>a priori</i> SNR uncertainty, soft mask estimator (SMPR) technique is selected for integration into the LPC-based speech coding system due to its superior performance in suppressing noise. The performance of the proposed speech coding system is then analyzed using various speech quality and intelligibility metrics. The proposed system shows enhanced performance across enormous noise types and SNR levels compared to the baseline LPC system.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating and integrating superior noise suppression algorithm into speech coding system

  • 期刊论文

  • 发表日期：

    2025年5月30日

查看全文
  • M. R. Prasad,
  • Manjunath B. Talawar,
  • N. Jagadisha,
  • Sharana Basavana Gowda

摘要

In wireless communication systems, the presence of noise significantly degrades the quality of transmitted speech signals. Our primary objective is to improve both the speech quality and intelligibility of coding systems in real-world noisy environments. This work presents an advanced approach for enhancing the robustness of linear predictive coding (LPC)-based speech coding system for various noise types and signal-to-noise ratio (SNR) levels. We begin by conducting a comprehensive evaluation of six noise suppression techniques to determine their efficacy using speech quality and intelligibility metrics. Following the evaluation, the a priori SNR uncertainty, soft mask estimator (SMPR) technique is selected for integration into the LPC-based speech coding system due to its superior performance in suppressing noise. The performance of the proposed speech coding system is then analyzed using various speech quality and intelligibility metrics. The proposed system shows enhanced performance across enormous noise types and SNR levels compared to the baseline LPC system.