DAGA 2010 - Deutsche Gesellschaft für Akustik eV

156 DAGA 2010 Programm
coefficients, and hence limit the available information for spectral envelope
estimation. In this paper, some new techniques based on cancellation
of rahmonics, rather than hard liftering, are proposed and examined
for their effectiveness in maintaining the spectral envelope information.
Both straightforward implementations and iterative procedures are considered
and simulation results for various configurations of F0 and formant
frequencies are presented. These preliminary examinations allow
the evaluation of effects of various acoustical and signal processing factors
on estimation accuracy and assess the feasibility of the proposed
approaches for use with high fundamental frequency signals, such as
singing, and in other similar fields of interest in musical acoustics.
Mi. 16:55 Grashof C 20 Signalverarbeitung
Super-Wideband Bandwidth Extension for Wideband Audio Codecs
Using Switched Spectral Replication and Pitch Synthesis
B. Geiser, H. Krüger und P. Vary
Institute of Comm. Systems and Data Proc. (IND), RWTH Aachen
This paper describes a new bandwidth extension algorithm which is targeted
at high quality audio communication over IP networks. The algorithm
is also part of the Huawei/ETRI candidate for the ITU-T superwideband
(SWB) extensions of Rec. G.729.1 and G.718. In the codec,
the 7-14 kHz frequency band of speech and audio signals is represented
in terms of temporal and spectral envelopes. This description is encoded
and transmitted to the decoder. In addition, the input signal’s fine
structure is analyzed and compactly encoded. From this compact information,
the decoder can regenerate the 7-14 kHz fine structure either
by spectral replication or by pitch synthesis. Then, an adaptive envelope
restoration procedure is employed. The algorithm operates in the MDCT
domain to allow subsequent refinement coding by vector quantization of
spectral coefficients. In the paper, relevant listening test results for the
G.729.1-SWB candidate codec that have been obtained during the ITU-
T standardization process are summarized. Good audio quality could be
shown for both speech and music signals.
Mi. 17:20 Grashof C 20 Signalverarbeitung
Methods for De-noising Impulse Response Measurements
X. Wang und M. Vorländer
Institut für Technische Akustik, RWTH Aachen
Impulse responses and transfer functions are the main properties of linear
acoustic systems. But during the measurements, the systems are
disturbed by external noise, (e.g. random noise, traffic noise or machine
noise); therefore the signal-to-noise ratio (SNR) is not sufficient. Generally,
there are two methods to de-noise: one is by making the excitation
signals such as MLS or sweeps longer; the other is to repeat and average
the sampled block periods. However, sometimes we can not make a
very long sweep, for instance, since the memory of a portable device is