頁籤選單縮合
題名 | 低速率即時語音編碼解碼器之設計與實作=Design and Implementaiton of a Low Data Rate LPC-based Real-time Speech Codec for Wireless Communication Applications |
---|---|
作者 | 劉家宏; 文星超; Liu, Chia-horng; Wen, Shing-chau; |
期刊 | 電信研究 |
出版日期 | 20030200 |
卷期 | 33:1 2003.02[民92.02] |
頁次 | 頁91-110 |
分類號 | 312.23 |
語文 | chi |
關鍵詞 | 語音編碼; 波形語音編碼法; 線性估計; 語音週期; 平均數值差函數法; 激發訊號; 正規方程式; 平均印象分數; Speech coding; Waveform coding; Linear prediction; Pitch; Average magnitude different function; Excitation signal; Normal equation; Mean opinion score; MOS; |
中文摘要 | 在通訊系統中所傳送的資料型態是以語音訊號為主。為有效利用有限的通訊頻道,語音訊號在傳送出去之前必須先經過語音壓縮技術的處理。尤其在無線通訊的應用上,更需要高壓縮比率的語音編碼方法,以提供低位元率的語音傳輸(低於8 kb/s)。傳統的波形語音編碼器受限於8 kHz的取樣頻率,所以其位元率必須大於8 kb/s。為了達到低位元率的目的,必須利用線性估計的技巧來進一步降低位元率,使其低於8 kb/s。在線性估計技巧的啟發下,許多根據這種語音編碼解碼架構所設計出來的新的語音編碼方法被提出來[1]。在本篇論文中,基於此一語音編碼解碼架構設計了一套位元率為3.5 kb/s的線性估計語音編碼解碼器,期能使用於現在的無線個人通訊系統(PCS)、軟體無線電通訊系統(SDR)、以及網際網路電話(Internet telephone, I-phone)等相關應用之中。在本篇論文中,將介紹基於線性估計語音編碼器的基本架構與工作原理。並藉由電腦模擬的驗證,比較經過此一編碼解碼器處理後的語音與原始語音,然後設計了語音編碼器的相關參數。由電腦模擬的結果可以知道,在3.5 kb/s的位元率下,線性估計語音編碼解碼器的平均印象分數大約是2.6左右,這個品質就相當於語音合成的品質。除此之外,我們透過聲霸卡進行即時語音的擷取,並經由此一線性估計語音編碼解碼器處理後,直接經由聲霸卡播放出來。如此,更可直接分析此一語音編碼解碼器的品質,並且使其具有可攜帶性(portable)與高靈活度(flexible)。為了讓使用者方便使用這一套即時語音編碼解碼器,我們也設計了一簡單的圖形化人機介面。 |
英文摘要 | Voice communication is the main application for a wireless communication system. In order to make use of the finite channel resource effectively, it is necessary to compress speech signal before the transmission. For the wireless communication, the speech coding schemes with high compression ratio is especially required. The data rate for these speech coding schemes is usually below 8 kb/s. The data use rate of traditional speech coding schemes using waveform coding technique is usually higher than 32 kb/s. Therefore, linear prediction based speech coding (LPC) algorithm are called for low data rate applications. In this paper, a LPC-based speech codec is designed with a data rate of 3.5 kb/s. The codec architecture and operation scheme are illustrated in the paper. The near optimum coefficients for the designed speech codec are also drawn by computer simulation. It is shown that MOS (Mean Opinion Score) of the designed speech codec is about 2.6. The issues of packet transmission are also discussed in the paper. Furthermore, a real-time implementation of the designed speech codec using sound blaster card is presented. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。