TH28599A - Speech encoding equipment and methods. - Google Patents

Abstract

The finished machine and speech encoding method, with this, allows the bit rate of the encoded data to be varied. The output vector x is quantized as a weighted vector by the first quantization unit of the 502 quantization unit of the 500 quantization unit. The shape index is output at output terminal 503, while the quantized value X0' is subtracted from the source vector x at adder 505. The quantized error vector y is dimensionally divided by the second quantization unit of the 510 quantization unit. The resulting quantized shape index data, which is weighted vector, is output at output terminals 5121, 5122. The quantized values Y1' and Y2' are dimensionally combined and integrated by adder 513 with the values. After quantizing X0', the resulting quantized value X1 will be output.

Claims (9)

1. The speech encoding method for dividing the input speech signal on the time axis into blocks, as units, and the resulting signal encoding consists of the following steps: finding the minimum short-term prediction resistance for the annoic portion of the input speech signal; determining the sine analysis encoding parameters based on the found short-term prediction error; and performing quantization vectors that give tangential weights to the sine analysis encoding parameters; and encoding the annoic portion of the input speech signal by waveform encoding. 2. The speech signal encoding method described in Reputation 1 involves determining whether the input speech signal is voiced or silent. Based on this determination, the voiced portions of the input speech signal found to be voiced are processed using sine analysis encoding, while the silent portions are quantized using a closed-loop vector-optimization search with synthesis analysis. 3. The speech signal encoding method described in Claim 1 uses spectral representations of the elopes as parameters for the analysis of the sine wave, which is then quantized with tactile weights. 4. A complete speech encoding machine for dividing the input speech signal on the time axis into blocks, such as units, and encoding the resulting signal, which consists of: a device for finding the minimum short-term prediction resistance of the input speech signal; a device for determining the parameters of sine analysis encoding based on the found short-term prediction resistance; a device for performing tangent weighted vector quantization on the sine analysis encoding parameters; and a device for encoding the silent portion of the input speech signal by waveform encoding. 5. The speech encoding method for dividing the input speech signal on the time axis into blocks, as units, and the resulting encoding signal consists of the following steps: finding the minimum short-term prediction resistance for the audio portion of the input speech signal; determining the parameters of the sine analysis encoding based on the found short-term prediction error; and performing tactile weighting vector quantization on the sine analysis encoding parameters. 6. The speech encoding procedure as provided for in Reputation 5 shall at least consist of: the first quantization vector step; and the second quantization step of the quantization vector of the quantization error produced at the time of the first quantization vector. 7. The speech encoding method is governed by claim 6, in which, for low bit rates, the output of the first quantize vector step is taken, and in which, for high bit rates, the output of the first quantize vector step and the output of the second quantize vector step are taken. 8. A speech encoding device for splitting the input speech signal on the time axis into blocks, such as units, and encoding the resulting signal, which consists of: a device for finding the short-term prediction resistance of the input speech signal; a device for determining the parameters of sine analysis encoding from the obtained short-term prediction error; and a device for performing multi-stage vector quantization that tacitly weights the parameters of sine analysis encoding. 9. A portable radio terminal device comprising; an amplification device for amplifying the input speech signal; an A/D converter for A/D conversion of the modulated signal; a speech encoder for encoding the output speech of the A/D converter; a path encoder for decoding the encoded channel; a modulation device for modulating the output of the path encoder; a D/A converter for D/A conversion of the modulated signal; and an amplifier for amplifying the signal from the D/A converter for feeding the amplified signal to the antenna; the speech encoder, which further comprises; a device for determining the short-range predictive reduction of the input speech signal; a device for determining the parameters of sine analysis encoding from the determined short-range predictive reduction; a device for performing tangent weighted vector quantization on the parameters of sine analysis encoding; and a device for encoding the input speech signal by Waveform encoding (9 claims, 3 pages, 15 images)