| 摘要 |
PURPOSE:To provide the speech encoding device which can synthesize a speech with frequency characteristics close to those of an input speech and to be improved in quantization efficiency. CONSTITUTION:An encoding part A encodes a fundamental frequency parameter corresponding to the fundamental frequency of the input speech by a pitch estimation unit 2. A voiced/voiceless decision unit 4 encodes a voiced/voiceless decision parameter made to correspond to the result of a decision on a voiced band or voiceless band in each of higher harmonic bands divided by a frequency band divider 3. A parameter quantizer 6 quantizes a spectrum envelope parameter made correspond to a spectrum envelope calculated by a spectrum envelope calculator 5. Those codes are multiplexed by a multiplexer 7 and sent to a decoding part B. The decoding part B separates the multiplexed signal sent from the encoding part A into codes by the parameters by a demultiplexer 11; and a pitch decoder 12 calculates the fundamental frequency, a voiced/ voiceless decision decoder 13 calculates a voiced/voiceless decision, and a spectrum envelope decoder 14 calculates a spectrum envelope. A spectrum amplitude estimation unit 15 finds the amplitude value of higher harmonics corresponding to the spectrum envelope. |
