Portable speakers or mobile devices with a small pair of speakers becomes increasingly popular. One of the major issues of the devices is that they emit a narrow and flat sound image like a mono sound source, although the device itself is stereo capable. Due to the compactness required from design trend, the speaker drivers are usually placed too close to each other thus makes the stereo content playback sounds lacks enough stereo wideness and timbre distortion due to crosstalk between the speakers. The general way of sound enhancement for this kind of situation uses sound pre/post-processing before it goes to the analog amplifier of the speaker. This type of processing usually includes boost out-of-phase signal and HRTF technology. However, this is never perfect as it causes un...[ Read more ]

Portable speakers or mobile devices with a small pair of speakers becomes increasingly popular. One of the major issues of the devices is that they emit a narrow and flat sound image like a mono sound source, although the device itself is stereo capable. Due to the compactness required from design trend, the speaker drivers are usually placed too close to each other thus makes the stereo content playback sounds lacks enough stereo wideness and timbre distortion due to crosstalk between the speakers. The general way of sound enhancement for this kind of situation uses sound pre/post-processing before it goes to the analog amplifier of the speaker. This type of processing usually includes boost out-of-phase signal and HRTF technology. However, this is never perfect as it causes unstable center front sound image for speech content and sensitive to individual’s body shape and not specifically optimized for to close pair speaker. This work proposed a method: to better enhance the sound stage immersiveness at the same time keeping an adjustable balance of foreground and background sound. The method is based on a modified RACE algorithm on top of that a Upmix algorithm, a HRTF renderer and a pseudo stereo technique. The key novelty of the method is to put source separation techniques on mono and stereo input and treat separated foreground and background sound independently for spatial enhancement and it has showed impressive performance in the subjective and objective evaluation compared to a third-party sound enhancement plug-in on the market.