ERROR CONFIGURATIONS IN BIT PATTERNS OF COMPRESSED DIGITAL SPEECH BASED ON CASCADED APPROACH

Abstract

The research challenges in digital speech processing remain in the traditionally identified area like speech compression. The main purpose to compress speech is to reduce the load of voice traffic for transmission, maintaining the quality of speech. In the initial stage of proposed research a cascaded approach is designed to reduce the bit rate and to the maintain speech quality. This coding technique is developed upon a PCM coded speech by dropping few of the least significant bits (LSB) of every byte and randomly substituting these dropped bits at the decoding end from their respective fixed Codebooks. To analyze the speech quality Mean Opinion Score is conducted, which is a subjective quality analysis. Experiments are conducted to refine the strategy by configuring the LSB bits after substitutions from their respective fixed codebooks but of reduced sizes. It is observed that as the codebooks containing the less number of bit patterns with lesser numeric values like 0 and 1, when substituted, their bytes have no changes in the valuable least significant bits.