| 摘要 |
PURPOSE:To prevent a pitch error from being generated by employing long- period prediction for the encoding of pitch synchronism by a speech synthesizer. CONSTITUTION:This speech encoding and decoding device is provided with an A/D conversion part 1 which obtains a digital speech signal from an analog signal, a synthesized waveform storage part 2 stored with a synthesized waveform obtained by decoding a last code, an acoustic analysis part 3 which finds a long-period prediction coefficient and a basic waveform of one pitch by taking a pitch analysis and a long-period predictive analysis of the digital speech signal, and a speech synthesis part 5 which generates the synthesized speech according to the codes of the acoustic analysis part 3 and basic waveform encoding part 4 and stores it in the synthesized waveform storage part 2. Then the device is provided with a basic waveform decoding part 7 which decodes the basic waveform of one pitch, a speech waveform decoding part 7 which decodes the digital speech signal, and an A/D conversion part 8 which converts the digital speech signal outputted by the speech waveform decoding part 7 into the analog speech signal. |
