4.2 Related techniques

4.2.1 Pseudostereophonic processes

BCC relies on a synthesis technique which can generate stereo and multi-channel signals given a mono signal. There is a long history of techniques attempting to ‘enhance’ mono signals to create a spatial impression, i.e. to generate a signal pair or more channels evoking some kind of spatial impression. Such techniques are often called ‘pseudostereophonic’ processes. Janovsky [151] proposed a scheme where a lowpass filtered version of the mono signal is given to one loudspeaker and a highpass filtered version to the other loudspeaker. Another technique uses complementary comb filters for generating left and right signals [179]. Schroeder [232] proposed the use of allpass filters instead of comb filters resulting in a stereo effect with less coloration artifacts. The use of a reverberation chamber with one loudspeaker emitting the mono signal and two microphones generating left and right signals was described by Schroeder [231] and Lochner and Keet [184]. Another scheme gives the mono signal to both loudspeakers and adds an attenuated and delayed version of the mono signal to one loudspeaker and the same phase-inverted attenuated and delayed signal to the other loudspeaker [179, 180]. Enkl [80] proposed the use of time-variant controllable filters controlled by properties of the mono signal. A more thorough review on these pseudosterephonic processes is given in [26].

In all these techniques, the spatial distribution of the auditory objects is independent of where the sound was originally picked up. The fundamental difference between these early techniques for ‘enhancing’ mono signals and the technique applied in BCC is that not an arbitrary auditory spatial image is to be created, but an auditory spatial image similar to the auditory spatial image of the original audio signal. For this purpose information about the auditory spatial image must be available. The before-mentioned ‘perceptually relevant differences’ between the audio channels represent this information.

4.2.2 Intensity stereo coding

Intensity stereo coding (ISC) is a joint-channel coding technique that is part of the ISO/IEC MPEG family of standards [32, 40, 136, 137, 249]. ISC is applied to reduce ‘perceptually irrelevant information’ of audio channel pairs and originated from [267]. In each coder sub-band that uses ISC, the sub-band signals are replaced by a down-mix signal and a direction angle (azimuth). The azimuth controls the intensity stereo position of the auditory object created at the decoder. Only one azimuth is transmitted for a scale factor band (1024 coder bands are divided into roughly 50 scale factor bands that are spaced proportionally to auditory critical bands). ISC is capable of significantly reducing the bitrate for stereo and multi-channel audio where it is used for channel pairs. However, its application is limited since intolerable distortions can occur if ISC is used for the full bandwidth or for audio signals with a highly dynamic and wide spatial image [119]. Potential improvements of ISC are constrained since the time–frequency resolution is given by the core audio coder and cannot be modified without adding considerable complexity.

Some of the limitations of ISC are overcome by BCC by using different filterbanks for coding of the audio waveform and for parametric coding of spatial cues [13]. Most audio coders use a modified discrete cosine transform (MDCT) [188] for coding of audio waveforms. The advantages of using a different filterbank for intensity stereo are reduced aliasing [13] and more flexibility, such as the ability to efficiently synthesize not only intensities (ICLD), but also time delays (ICTD) and coherence (ICC) between the audio channels. Another notable conceptual difference between ISC and BCC is that the latter operates on the fullband audio signal and transmits only a mono time domain signal, whereas ISC does not transmit a true mono fullband signal, but only operates on a certain (high) frequency range.