TH76751B - Speech Speech Encoder, Speech Decoder, Speech Recoding and Speech Decoding Method - Google Patents

Abstract

New Request Update 12/04/2016 Speech-to-speech encoder, Speech-to-speech decoder, Speech-to-speech encoding method Speech and phonological decoding methods, designed to improve sound quality, by performing sharing effectively bit parts Dominant Frequency Band Distinguisher (301), performs a dominant frequency band distinction. has the greatest norm coefficient in the input speech spectrum, the dominant group (302-1 - 302-N) and the non-dominant (303) grouping factor. bar All subfrequency is the dominant group with the dominant band included, and the non-dominant group that does not have the dominant band included. Group bit allocation (308), does the bit allocation of each group, based on power and Norm distribution of each group, subband bit allocation (309), allocated bits that are allocated to each subband each longer in proportion to the norm. to the energy of the group

Claims (8)

Disclaimer (all) which will not appear on the advertisement page. : New request updated 12/04/2016 1. A speech audio encoder consisting of, a transducer converts the input signal from time to band, and, the projection path, estimates the energy shape frame, for each subband which is divided. The frequency spectrum of the input signal and, the quantization path quantifies the said energy framing and the group determines the frame of such energy that is quantized into multiple groups. And, 1st bit allocation path allocates bits at the aforementioned groups and, 2nd bit allocation path allocates bits allocated to that multiple groups at the bar. Subfrequency that each group and, encoded path encodes the spectrum of that frequency, using the allocated bits at that subband where the path determines the group, the dominant frequency band is decided and, the subband that Is a downward inclination of The energy contours on both sides of the dominant frequency band form the dominant group, cutting successive sub-bands other than the dominant frequency band into inconspicuous segments. 2. Speech-coding equipment specified in claim 1, which includes, the energy-finding path, calculates the power of each group, and the decentralized path calculates the frame-dispersion. Shape of power The work of each group, where the 1st-order bit-allocated pathway, references the energy of each of these groups and the scattering of the energy of each group. Each of these groups calculated, the greater at least on one side of the energy and dispersion. The greater the energy geometries, the more bits are allocated per group, the more at least. On the one hand, energy and energy dispersion is less, the greater the allocation. Fewer bits per group 3. Speech encoding device specified in claim 1 where, the 2nd bit allocation pathway, the greater the energy frame of such subband, the more bits are allocated. number The greater the frequency per this subband, the smaller the energy contour of the subband, the fewer bits are allocated than this subband. 4. Speech-based audio decoding device, which consists of, the reverse quantization path, reverses the quantization of the quantized spectrum, and, the group determines the contour of the spectrum. And, 1st order bit allocation pathways allocates bits at such multiple groups and, 2nd bit allocates bits allocated bits at such multiple groups. At the bar The subfrequency at each group and, the decoding path, decodes the frequency spectrum of the speech audio contract, using the bits allocated to that subband and, the path up the frame shape makes the decoded spectrum reappear, by applying Picture frame A quantized figure of such spectrum is reversed at the decoded frequency spectrum and, the reverse transformed path, reverses the decoded spectrum from the frequency band to the time range at which the path determines the segment. Said, judging the dominant frequency band and, the slanted subband of The energy contour frames on both sides of the dominant frequency band form the dominant group, making the judgment of the sub-continuation bands other than the dominant frequency band as the dominant group. 5. Speech decoding equipment identified in claim 4, which includes additionally, the energy calculation path, calculates the energy of each group and, the decomposition path, calculates the scatter distribution. Of power The work of each group, the aforementioned bit-bit allocation pathway, references the energy of each of these groups and the scattering of the energy of each group. Each of these groups calculated, the greater at least on one side of the energy and dispersion. The greater the energy geometries, the more bits are allocated per group, the more at least. On the one hand, energy and energy dispersion is less, the greater the allocation. Fewer bits per group 6. Speech decoding equipment specified in claim 4 where, the second bit allocation pathway, the greater the energy frame of the subband, the more bits are allocated. The greater the frequency per this subband, the smaller the energy contour of the subband, the fewer bits are allocated than this subband. 7. Speech coding method where, in which, converts the input signal from time to band, estimates the energy shape frame, for each frequency band. The multi-band subdivision which divides the frequency spectrum of the input signal, quantifies the said energy scaffold, gives the quantization of that energy into multiple groups, allocates bits at multiple groups. Such groups, allocate bits allocated to the aforementioned sub-band at each subband, encode that frequency spectrum, using bits allocated at each subband of each group, 8. Speech-based audio decoding method where, in which, the quantization reverses the quantized spectrum of the quantized spectrum, makes the quantization of such spectrum framing multiple groups, allocates bits. At such multiple groups, allocated bits are allocated to the aforementioned sub-bands at each subband, decoding of the frequency spectrum of the speech audio signal, using bits allocated to the band. Such subfrequency, revives the decoded spectrum, by applying the spectral framework as Inverted quantization at the decoded frequency spectrum, reverse conversion of the decoded spectrum from frequency band to time band is performed.