Binaural Unmasking in Practical Use: Perceived Level of Phase-inverted Speech in Environmental Noise

We aim to develop a technology that makes the sound from earphones and headphones easier to hear without increasing the sound pressure or eliminating ambient noise. To this end, we focus on harnessing the phenomenon of binaural unmasking through phase reversal in one ear. Specifically, we conduct experiments to evaluate the improvement of audibility caused by the phenomenon, using conditions that approximate practical scenarios. We use speech sounds by various speakers and noises that can be encountered in daily life (urban environmental sounds, cheers) to verify the effects of binaural unmasking under conditions close to practical situations. The results of experiments using the Japanese language showed that (i) speech in a noisy environment is perceived to be up to about 6 dB louder with phase reversal in one ear, and (ii) a certain effect (improvement of audibility by 5 dB or more) is obtained for all speakers and noises targeted in this study. These findings demonstrate the effectiveness of binaural unmasking attributed to interaural phase differences in practical scenarios.

💡 Research Summary

The paper addresses a common dilemma in headphone and ear‑phone usage: users want to hear target audio (e.g., music, speech) more clearly while still being aware of ambient sounds such as traffic or announcements. Conventional solutions—passive or active noise cancellation and speech‑enhancement algorithms—either block or heavily process the surrounding sound, which conflicts with the desire to retain environmental awareness. The authors propose exploiting the well‑known psychoacoustic phenomenon of binaural unmasking, specifically by introducing a 180‑degree phase inversion of the target speech in only one ear. This creates an interaural phase difference (IPD) that the auditory system can use to separate the target from the masker, thereby improving perceived loudness without increasing actual sound pressure level.

To move beyond prior laboratory studies that typically used a single male speaker and synthetic noises (white or speech‑shaped noise), the authors designed an experiment that more closely mirrors real‑world listening conditions. Three speech recordings were used: one male and two female voice actors, each delivering the same Japanese sentence from the ITA corpus. The female voices were selected because they exhibit distinct formant patterns, ensuring a broader representation of vocal characteristics. Three masker noises were also employed: (A) white noise generated in MATLAB, (B) a recorded crowd cheer, and (C) a composite of urban environmental sounds (traffic, construction, etc.). These noises were chosen because they are commonly encountered in everyday life and cannot be completely eliminated for safety or communication reasons.

The core contribution is the introduction of a new metric, the Binaural Hearing Level Difference (BHLD). While traditional Binaural Masking Level Difference (BMLD) measures the reduction in detection threshold, BHLD quantifies the change in perceived loudness when listeners adjust the level of the original (non‑inverted) speech until it matches the perceived loudness of the phase‑inverted version, both presented with the same masker. Participants performed this adjustment in 1‑dB steps using a MATLAB‑based graphical interface that displayed only numeric volume values, preventing bias from visual cues. The volume of the masker remained constant; only the speech level was altered.

Sixteen native‑Japanese participants (11 male, 5 female, ages 20‑59) were recruited from a corporate pool, excluding anyone with self‑reported hearing problems. Each participant completed a single session in which they compared the original (S₀) and phase‑inverted (Sπ) speech for each combination of speaker and masker. One participant’s data were discarded due to inconsistent responses, leaving 16 valid data sets.

Results consistently showed a positive BHLD across all speaker‑masker pairs. The average improvement was about 5 dB, with a maximum of roughly 6 dB. The effect was most pronounced in the low‑frequency band (200–500 Hz), aligning with earlier physiological findings that the superior olivary complex is highly sensitive to interaural time differences at low frequencies. At higher frequencies, the benefit tapered to around 3 dB but remained measurable up to 4 kHz. Importantly, the improvement was observed for both male and female voices, indicating that the phenomenon is robust across different vocal formant structures.

The authors discuss several implications. First, a 5 dB increase in perceived level can shift a listening scenario from “barely audible” to “comfortably audible” without raising the actual SPL, thereby reducing the risk of hearing‑damage associated with high playback volumes. Second, because the benefit is concentrated in the low‑frequency region, a practical implementation could apply phase inversion selectively to that band, minimizing computational load and power consumption in portable devices. Third, the BHLD metric provides a user‑centric measure that directly reflects perceived audibility, which may be more relevant for product development than traditional detection‑threshold metrics.

Limitations include the reliance on self‑reported hearing status rather than audiometric screening, the static laboratory setting (participants seated, using wired headphones), and the fact that only perceived loudness—not speech intelligibility—was evaluated. The authors acknowledge that intelligibility, especially for consonant‑rich high‑frequency content, may require additional processing.

Future work is outlined as follows: (1) integration of real‑time, low‑latency phase‑inversion algorithms into open‑ear ear‑phones; (2) expansion of the participant pool to include non‑Japanese speakers and a broader age range; (3) objective audiometric testing to correlate BHLD with individual hearing thresholds; (4) assessment of long‑term listening comfort and potential fatigue; and (5) exploration of hybrid approaches that combine selective phase inversion with modest gain control to further boost intelligibility.

In conclusion, the study provides experimental evidence that binaural unmasking via unilateral phase inversion can yield a perceptual gain of up to 6 dB in realistic listening environments. This gain is achieved without increasing overall sound pressure or suppressing ambient noise, making the technique a promising candidate for next‑generation audio devices that aim to enhance target speech while preserving situational awareness.