摘要:
Apparatus for simultaneously decompressing and interpolating compressed audio data. The compressed audio data is stored in differential log format, meaning that the difference between each two consecutive data points is taken and the log of the difference calculated to form each compressed data point. To efficiently decompress and interpolate the compressed data, advantage is taken of the fact that addition of logs is equivalent to multiplication of linear values. Thus the log of an interpolation factor is added to each compressed data point prior to taking the inverse log of the sum. An integrator block completes the interpolation and decompression of the data.
摘要:
A model adaptation scheme in the pattern recognition, which is capable of realizing a fast, real time model adaptation and improving the recognition performance. This model adaptation scheme determines a change in a parameter expressing a condition of pattern recognition and probabilistic model training between an initial condition at a time of acquiring training data used in obtaining a model parameter of each probabilistic model and a current condition at a time of actual recognition. Then, the probabilistic models are adapted by obtaining a model parameter after a condition change by updating a model parameter before a condition change according to the determined change, when the initial condition and the current condition are mismatching. The adaptation processing uses a Taylor expansion expressing a change in the model parameter in terms of a change in the parameter expressing the condition.
摘要:
A signal encoding apparatus for encoding an acoustic signal. This signal encoding apparatus includes a transform circuit for transforming an inputted acoustic signal into frequency components, a signal component separating circuit for separating an output of the transform circuit into tone characteristic components and noise characteristic components, a tone characteristic encoding circuit for encoding a signal of tone characteristic components, and a noise characteristic component encoding circuit for encoding a signal of noise characteristic components, wherein the tone characteristic component encoding circuit encodes respective signal components of the signal of tone characteristic components so that they respectively have different code lengths to thereby improve efficiency of encoding without degrading sound quality with respect to acoustic signal of tone characteristic.
摘要:
A stimulus waveform is processed using a model of the human auditory system to provide a plurality of output waveforms. Each output waveform corresponds to excitation at different locations along the basilar membrane in the cochlea, and matches the narrow frequency bandwidth, short time response, and wave propagation characteristics of the human cochlea. Primary feature detection is achieved by comparing response waveforms and their spatial and time derivatives to predetermined stereotypes. Secondary feature detection is achieved by comparing spatial and temporal patterns of primary features with patterns stereotypical of human speech elements.
摘要:
An efficient implementation of oddly-stacked critically-sampled single sideband analysis/synthesis filter banks is achieved by application of a set of functions to time-domain and frequency-domain values before and after transformation. In one embodiment of an analysis filter bank, a forward pre-transform function groups blocks of N samples into blocks of 1/4N modified samples, a discrete transform generates frequency-domain coefficients in response to the modified samples, and a forward post-transform function generates spectral information in response to the frequency-domain transform coefficients. In one embodiment of a synthesis filter bank, an inverse pre-transform function groups spectral information into blocks of 1/4N frequency-domain transform coefficients, a discrete transform generates blocks of 1/4N time-domain transform coefficients in response to the frequency-domain transform coefficients, and an inverse post-transform function generates blocks of N time-domain samples in response to the time-domain transform coefficients. An implementation of an oddly-stacked Time Domain Aliasing Cancellation transform permits the length of the transformation to be adaptively selected.
摘要翻译:通过在变换前后对时域和频域值应用一组函数来实现奇数堆叠临界采样单边带分析/合成滤波器组的有效实现。 在分析滤波器组的一个实施例中,正向预变换功能将N个采样的块分组为+ E,fra 1/4 + EE N个修改采样的块,离散变换响应于修改的采样生成频域系数 并且前向后变换功能响应于频域变换系数产生频谱信息。 在合成滤波器组的一个实施例中,逆预变换功能将频谱信息分组为+ E,fra 1/4 + EE N个频域变换系数的块,离散变换生成+ E, + EE N时域变换系数,并且逆后变换函数响应于时域变换系数生成N个时域样本的块。 奇异堆叠的时域混叠取消变换的实现允许自适应地选择变换的长度。
摘要:
Transmitter, encoding system and method for subband coding a digital signal. The encoding system includes a splitter unit for dividing the digital signal into subband signals SB.sub.1, . . . , SB.sub.p ; a quantizer unit for quantizing time-equivalent q sample signal blocks of the subband signals; a bit need determining unit and a bit allocation unit. The bit need determining unit determines a bit need b.sub.m which corresponds to the number of bits by which the q samples in a time-equivalent signal block in a subband signal SB.sub.m should be represented, where 1.ltoreq.m.ltoreq.P. The bit allocation unit allocates n.sub.m bits to each of the q samples of the time-equivalent signal block of subband signal SB.sub.m on the basis of the bit need b.sub.m and an available bit quantity B, n.sub.m being the number of bits by which the q samples in the time-equivalent signal block of subband signal SB.sub.m will be represented, where 1.ltoreq.m.ltoreq.P.
摘要:
A method of passing tones transparently over narrowband channels without having to detect the tones at the transmit side is described. This method requires neither separate handling of tones for the transmission purpose, nor prior knowledge of the characteristics of the signaling tones used. It is a relatively general method that can be applied to systems using a variety of speech coding techniques utilizing a frame structure.
摘要:
A sinusoidal model for acoustic waveforms is applied to develop a new analysis/synthesis technique which characterizes a waveform by the amplitudes, frequencies, and phases of component sine waves. These parameters are estimated from a short-time Fourier transform. Rapid changes in the highly-resolved spectral components are tracked using the concept of "birth" and "death" of the underlying sine waves. The component values are interpolated from one frame to the next to yield a representation that is applied to a sine wave generator. The resulting synthetic waveform preserves the general waveform shape and is perceptually indistinguishable from the original. Furthermore, in the presence of noise the perceptual characteristics of the waveform as well as the noise are maintained. The method and devices are particularly useful in speech coding, time-scale modification, frequency scale modification and pitch modification.
摘要:
Respective samples of input speech data are transformed by a discrete Fourier transformer to obtain a spectrum of the speech data. Simultaneously, values of times of the respective samples are multiplied with the input speech data by a multiplier and a differential spectrum is obtained by transforming a result of the multiplication by a discrete Fourier transformer. A real part of a value obtained by dividing the differential spectrum by the spectrum by a quotient real part calculator and the real part is inverse-transformed by a discrete inverse Fourier transformer. The result of the inverse-transformation is divided by the values of the times of the respective samples to obtain a time function corresponding to phase. On the other hand, a time function corresponding to a logarithmic amplitude spectrum is obtained from an output of the inverse-Fourier transformer by means of a logarithmic amplitude spectrum calculator. A complex cepstrum is obtained by adding the both time functions at respective times by an adder.
摘要:
An audio type signal is encoded. The signal is first divided into bands. For each band, a yardstick signal element is selected. Its magnitude is quantized using a first level of accuracy. This magnitude is used for various purposes, including assigning bits to the different bands, and for establishing reconstruction levels within a band. The magnitude of non yardstick signal elements is quantized with less accuracy than are the yardstick signal elements. The encoded signal is also decoded. Apparatus for both encoding and decoding are also disclosed. The location of the yardstick element within its band may also be recorded and encoded, and used for efficiently allocating bits to non-yardstick signal elements.