Dolby elicits multichannel sound from stereo headphones.
Multichannel surround sound is becoming increasingly important, not only for movie soundtracks but for music recordings. Of course, you need a playback system with at least six channels of amplification and corresponding speakers to hear these multichannel music disks and movie soundtracks in their full glory. But what if you want to listen to such a disc while your spouse sleeps in the next room - or even next to you in bed? With a stereo source, headphones are the obvious answer. But how can stereo headphones deliver multichannel sound?
Several companies have tried to address this problem without much success. I've listened to some of those companies' systems and been quite underwhelmed - until now. Lake Technology (www.laketechnology.com), an Australian R&D firm, has developed a sophisticated DSP algorithm that simulates the surround-sound experience astonishingly well, using any pair of stereo headphones. Lake has licensed the technology to Dolby Labs (www.dolby.com), which in turn licenses it to manufacturers under the name Dolby Headphone.
I first heard a Dolby Headphone demonstration at the 2000 Audio Engineering Society convention in Los Angeles. A small home-theater system was set up with five speakers and a subwoofer, in addition to a set of headphones for each seat. A short video presentation explained the system's basic principles and played various audio clips in stereo and multichannel mode for comparison. During the first multichannel clip, I took off the headphones to see if the external speakers were on; at that very moment, the pre-recorded announcer said, "No, the speakers aren't on!" The rich, open sound seemed to come from all around outside my head. In contrast, the stereo version seemed quite harsh and strident inside my head.
How can the multichannel experience be simulated with conventional stereo headphones? The Dolby Headphone algorithm models the acoustic behavior of five speakers and a listener in a room. It includes several primary elements: inter-ear arrival time (the difference between the moments at which a sound arrives at each ear), head-related transfer functions or HRTFs (which describe how a sound's spectrum changes as it diffracts around your head), and room response, including early reflections and reverb tails (see Fig. 1).
The acoustic path from each virtual speaker to each ear is simulated by a 7,000-tap finite impulse response (FIR) filter, which is said to be 350 times more precise than those used in previous headphone-surround systems. Ten such filters are required to reproduce the effect of five speakers placed around the listener. That many large filters normally require excessive processing power and introduce pronounced latency; Lake's new convolution techniques eliminate these problems.
Lake has coded the algorithm for several general-purpose DSP chips, including the Zoran ZR38650, Analog Devices ADSST-MELODY-1000, and Motorola DSP56300 family. In addition, the company has developed core code for application-specific integrated circuits (ASICs) dedicated to Dolby Headphone, which Sanyo Semiconductor recently announced it will manufacture. Finally, the algorithm can run as software on a PC with an Intel Pentium II or III or an AMD K6-II CPU running at a clock speed of at least 400 MHz.
The obvious application is to include a Dolby Headphone switch on consumer A/V receivers and surround processors; hit the switch, and the five main channels are routed through the algorithm to the headphone output. (Because bass frequencies are not directional, the ".1" LFE channel is passed unaltered to both sides of the headphones.) Other potential applications include game consoles and computers.
Dolby Headphone is set to become an important part of the emerging multichannel-audio market. I look forward to using it while my wife slumbers beside me.