We describe a novel method for designing echoic headphone virtualizers based on a stochastic room model and a numerical optimization procedure. The method aims to maximize sound source externalization under a natural-timbre constraint. The stochastic room model generates a number of binaural room impulse response (BRIR) candidates for each virtual channel, each embodying essential perceptual cues. A perceptually-based distortion metric evaluates the timbre of each candidate, and the optimal candidate is selected for use in the virtualizer. We designed a 7.1.4 channel virtualizer and evaluated it relative to a LoRo stereo downmix using a single-interval A:B preference test. For a pool of 10 listeners, the test resulted in an overall virtualizer preference of 75%, with no stereo test item preferred over binaural.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.