This paper presents an improved 2.4 kbps class-dependent CELP speech coder. The improved coder is based on our previous efforts on a 2.4 kbps CELP coder. New classification schemes in both open-loop and closed-loop ar...
详细信息
This paper presents an improved 2.4 kbps class-dependent CELP speech coder. The improved coder is based on our previous efforts on a 2.4 kbps CELP coder. New classification schemes in both open-loop and closed-loop are used. An extra transition class is used in the closed-loop classification. We developed a predictive LSP VQ to reduce LPC bit rate to as low as 18 bits/frame (30 ms) while maintaining low spectral distortion. Experimental results show that the quality of synthesized speech is improved. We also discuss the results obtained from class-dependent weighting filters, and a class-dependent postfilter.
This paper describes PSEUDEC, a dedicated co-processor and the rationale behind its design. The final goal of our work is to present a single chip solution with low power consumption for an advanced digital hearing ai...
详细信息
This paper describes PSEUDEC, a dedicated co-processor and the rationale behind its design. The final goal of our work is to present a single chip solution with low power consumption for an advanced digital hearing aid based on a parameterized transformation of speech (PARTRAN). Characterization of the constituent parts of the PARTRAN algorithm shows that it is well suited for implementation on a heterogeneous architecture. The design strategy used identifies a subset for implementation on dedicated hardware, with a computational complexity roughly equivalent to the performance of a standard 10 MIPS DSP. The subset of PARTRAN implemented by PSEUDEC performs pseudo-decomposition of a 12th order LPC polynomial. An adapted algorithm displays improved dynamic range compared to the conventional solution for DSPs, calculating the amplitude spectrum rather than the power spectrum. Highly pipelined CORDIC-units optimized for the application replaces complex multiplication, trigonometric operations (for e/sup j/spl omega//) and square root (for |a|=/spl radic/(a/sub r//sup 2/+a/sub i//sup 2/)), exploiting the power of CORDIC operations in advanced DSP algorithms. PSEUDEC uses redundant data representation and bit-serial arithmetic, most significant digit first (on-line arithmetic) for efficient implementation of operators and for efficient (inter-operator) communication. The inherent nature of on-line arithmetic and the operators used allows for fast and efficient implementation even when using ordinary standard cells.
An improved design for a transform-domain weighted interleave vector quantization (Twin VQ) audio-codec is proposed that is applicable to personal communication systems. Twin VQ is a low-bit-rate audio-coding method t...
详细信息
An improved design for a transform-domain weighted interleave vector quantization (Twin VQ) audio-codec is proposed that is applicable to personal communication systems. Twin VQ is a low-bit-rate audio-coding method that achieves high-quality under error-free conditions, at less than 64 kbit/s. To make this method applicable to wireless systems, which have a higher bit-error rate due to inferior transfer conditions, the Twin VQ coding algorithm has been improved. About 90% of Twin VQ encoded bits were found to be insensitive to channel errors, so using the BS-FEC method, error protection is weighted to the sensitive bits. Subjective evaluation tests showed that Twin VQ with BS-FEC maintains good quality over, high bit-error-rate channels.
This paper presents an algebraic vector-quantization scheme for encoding the LSF parameters used in describing the time-varying short-term spectrum of speech in many modern vocoders. The quantizer achieves an average ...
详细信息
This paper presents an algebraic vector-quantization scheme for encoding the LSF parameters used in describing the time-varying short-term spectrum of speech in many modern vocoders. The quantizer achieves an average spectral distortion of 1 dB at 28 bits/frame for the telephone bandwidth. The scheme is based on low-dimensionality regular-point lattices. The properties of lattices are taken advantage of in both the design and the search of the quantizer codebook. Namely, this algebraic codebook need not be stored in memory and the optimum vector is found through simple rounding of the input variables instead of the usual exhaustive search. Thus, the scheme results in significant savings of memory and reduced computational complexity when compared to traditional vector-quantizer solutions.
We describe four new methods for detection of cholesterol plaque deposits inside arteries, using digital signal processing. These methods have been tested on a number of patients. The results obtained are presented an...
详细信息
We describe four new methods for detection of cholesterol plaque deposits inside arteries, using digital signal processing. These methods have been tested on a number of patients. The results obtained are presented and discussed. An ultrasonic beam is directed towards the examined artery. Signal features obtained from the Doppler signal reflected from the bloodstream contain information about the character of turbulences caused by cholesterol deposits inside the artery. These features can thus provide information about the size, shape, and position of stenosis. Features in both the time and frequency domains were utilized, such as the LPC cepstrum, the exponential time-frequency distribution, the energy waveform and the wavelet transform.
暂无评论