Parametric Coding of Binaural Audio
Ladda ner
Publicerad
Författare
Typ
Examensarbete för masterexamen
Program
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Binaural cue coding, BCC, presumes that the signals in the two channels differ
primarily with respect to time, level, and coherence. The method then assumes
that both channels can be synthesised by imposing the according time, level, and
coherence differences onto a single-channel signal.
BCC can generally be split into two major parts: encoding and decoding. Encoding
represents the analysis of the signal and investigates the signal’s inter-channel time
and level differences as well as the inter-channel coherence. The signal’s channels are
then summed, the signal is downmixed. The decoding process upmixes the singlechannel
signal to its original number of channels and restores the inter-channel time
and level differences and uses the inter-channel coherence to mimic the inter-channel
coherence of the original audio signal.
Binaural signals, signals consisting of two channels each filtered with head related
transfer functions, HRTFs, are sensitive to compression algorithms since they depend
on a highly delicate time and level differences between the two channels. Thus, the
thesis is focused on the low-complexity implementation of the BCC scheme presented
by Faller in his Phd thesis Parametric Coding of Spatial Audio and how well this
scheme can synthesise the spatial cues present in binaural signals. The result and
analysis is divided into two parts: a simple analysis used to investigate the general
performance of the implemented algorithm and an informal listening test evaluating
how well the program handles the synthesis of the binaural signals.
The synthesised audio exhibits a varied intensity of artefacts if the ICTD synthesis is
incorporated in the BCC process. If the ICTD synthesis is left out of the BCC process,
almost no artefacts can be perceived. However, this reduces the lateralization
of the sounds/sound sources present in the audio. The conclusion that follow is thus
that the implementation illustrates the potential the method has for transferring
high complexity audio, such as binaural signals, at low bitrates.
Beskrivning
Ämne/nyckelord
BCC, binaural, ICC, ICTD, ICLD, audio, sound