• The Journal of the Acoustical Society of Korea
• /
• v.20 no.3
• /
• pp.34-41
• /
• 2001

In an IMBE (Improved Multi-band Excitation) speech coder, initial pitch estimation occupies most of the total computing time for the coder due to complex cost function and exhaustive search over candidate pitches. Future frames in initial pitch estimation cause inevitable time delay. Therefore, it is difficult to implement a real-time coder. Furthermore, unvoiced frames use the unnecessary pitch estimation as in the voiced frames. In this paper, each frame is determined voiced or unvoiced by Dyadic Wavelet Transform (DyWT) and, then, initial pitch estimation is performed only for voiced frame. Therefore different pitch estimation algorithms are employed between voiced and unvoiced frames incurring reduced time delay at transmitter and receiver. Simulation result show that the relative complexity of initial pitch estimation is reduced by 23％, and the processing time decreases down to 1/10 ∼ 1/1l of the IMBE coder while speech quality is almost maintained.

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.3E
• /
• pp.53-57
• /
• 1998

Spectral information at a speech coder should be quantized with sufficient accuracy to keep perceptually transparent output speech. Spectral information at a low bit rate speech coder is usually transformed into corresponding line spectrum pair parameters and is often quantized with a vector quantization algorithm. As the vector quantization algorithm generally has high complexity in the optimal code vector searching routine, the complexity reduction in that routine is investigated using the ordering property of the line spectrum pair. When the proposed complexity reduction algorithm is applied to the well-known split vector quantization algorithm, the 46% complexity reduction is achieved in the distortion measure compu-tation.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.3E
• /
• pp.3-6
• /
• 1997

This paper reports on the optimal design of the excitation codebook in the 13 kbps variable rate QCELP coder of Korean speech. We present two optimal excitation codebooks which consist of 128 and 556 samples, respectively. For the design and test of the improved codebook, a data base of Korean speech is used. A quasi-Newton optimization algorithm was developed to design the codebook. The optimized codebook which remains sparse, can produce an average gain of 0.84 and 0.45 dB in SNR and SEGSNR respectively. Informal listening tests confirm the improvement in speech quality.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.5
• /
• pp.35-47
• /
• 2001

This paper describes an efficient harmonic-CELP speech coder by taking advantages of harmonic and CELP coders into account. According to frame voicing decision, the proposed harmonic-CELP coder adopts the RP-VSELP coder as a fast CELP in case of an unvoiced frame, or an improved harmonic coder in case of a voiced frame. The proposed coder has main features as follows: simple pitch detection, fast harmonic estimation, variable dimension harmonic vector quantization, perceptual weighting reflecting frequency resolution, fast harmonic synthesis, naturalness control using band voicing, and multi-mode. These features make the proposed coder require very low complexity, compared with HVXC coder To demonstrate the performance of the proposed coder, a 2.4 kbps coder has been implemented and compared with reference coders. From results of informal listening tests, the proposed coder showed good quality while requiring low delay and complexity.

• Proceedings of the KSPS conference
• /
• 2002.11a
• /
• pp.175-178
• /
• 2002

Most existing telephone networks transmit narrowband speech witch has been bandlimited below 4 kHz. Compared with wideband speech up to 8 kHz, narrowband speech shows reduced intelligibility and a muffled quality. Bandwidth extension is a technique to generate wideband speech by reconstructing 4-8 kHz highband speech without any additional information. This paper presents experimental results of the bandwidth extension adopted for 4800 bps CELP speech coder. In this experiment, we examine various methods for reconstruction of wideband spectrum and excitation signal, compare and analyze their performance by performing the subjective preference test and measuring the cepstral distortion.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.4
• /
• pp.45-51
• /
• 2000

This paper describes real-time implementation of a multi-channel G.729A speech coder using a 16 bit fixed-point Digital Signal Processor (DSP) and its application to a Voice Mailing Service (VMS) system. TMS320C549 by Texas Instruments was used as a fixed point DSP chip and a 4 channel G.729A coder was implemented on the chip. The implemented coder required 14.5 MIPS for the encoder and 3.6 MIPS for the decoder at each channel. In addition, memories required by the coder were 9.88K words and 1.69K words for code and data sections, respectively. As a result, the developed VMS system that accommodates two DSP chips was able to support totally 8 channels. Experimental results showed that the our multi-channel coder passes all of test vectors provided by ITU-T.

• Speech Sciences
• /
• v.11 no.2
• /
• pp.259-269
• /
• 2004

In this paper, we propose a new excitation enhancement technique to improve the speech quality of low bit-rate code-excited linear prediction (CELP) coders. The proposed technique is based on a harmonic model and it is employed only in the decoding process of speech coders without any additional bits. We develop the procedure of harmonic model parameter estimation and harmonic generation, and apply this technique to a current state-of-the-art low bit rate speech coder, ITU-T G.729 Annex D. Also, its performance is measured by using the ITU-T P.862 PESQ score and compared to those of the phase dispersion filter and the long-term postfilter applied to the decoded excitation. It is shown that the proposed excitation enhancement technique can improve the quality of decoded speech and provide better quality for male speech than other techniques.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.2
• /
• pp.129-134
• /
• 2003

In this paper, we efficiently implement the AMR speech coder by reducing the complexity of algebraic codebook search. To reduce the computational complexity of the algebraic codebook search, we propose a fast algebraic codebook search method that improves conventional depth first tree search method used in AMR speech coder algorithm. The proposed method reduces the search complexity by pruning the trees which are less possible to be selected as an optimum excitation. This method needs no additional computation for selecting the trees to be pruned and reduces the computational complexity considerably compared to the original depth first tree search method with slightly degradation or speech qualify. Applying our method to the implementation or AMR speech coder with 12.2 kbps mode by using the TeakLite DSP, we reduce the search complexity about 40% compared to the conventional method.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.51-54
• /
• 2000

This paper describes the fixed-point DSP implementation of a CELP(code-excited linear prediction)-based speech coder. The effective realization methodologies to maximize the utilization of the DSP's architectural features, specifically Parallel movement and pipelining are also presented together with the implementation results targeted for the ITU-T standard G.723.1 using Motorola DSP56309. The operation of the implemented speech coder is verified using the test vectors offered by the standard as well as using the peripheral interface circuits designed for the coder's real-time operation.

• Proceedings of the IEEK Conference
• /
• 2001.06d
• /
• pp.115-118
• /
• 2001

This paper presents dispersed-pulse and random codebook for CELP coder. This coder operates on speech frames of 20ms and generates an excitation vector by convoluting dispersion vectors with signed pulses in an algebraic codevector. The improvement of pulse-based fixed codebook is performed at a low bit rate. A high performance fixed-codebook consists of a partial algebraic codebook and a random codebook in unvoiced and stationary noise regions. The proposed CELP coder is quantized with 4kb/s and is compared with G.729 (Bkb/s CS-ACELP). Subjective testing shows better quality than reference coders under some background noise conditions