Recent advancements in headphone and earphone personalization have significantly improved spatial audio by tailoring binaural rendering to individual listeners. These methods typically focus on accurately reproducing spatial attributes such as localization and envelopment. However, the equally important aspect of natural timbre reproductionespecially for conventional stereophonic contenthas been largely overlooked. Most existing personalization techniques are designed for immersive formats and do not adequately address the perceptual quality of timbre when listening through earphones.
This paper presents a novel personalization method aimed at enhancing the perception of natural timbre in stereophonic sound via earphones. Unlike traditional EQ-based approaches that adjust frequency responses based on user preferences or population averages, our method is grounded in an original auditory model. This model isolates and optimizes perceptual cues directly related to timbre recognition, independently of spatial attributes, thereby improving
timbral naturalness without compromising localization or stereo imaging.
The proposed method consists of four key stages. First, a high-resolution 3D scan of the listeners upper body, including detailed pinna geometry, is acquired. Second, individualized Head-Related Transfer Functions (HRTFs) are computed using the boundary element method for 864 sound incidence directions. In the third stage, spatial encoding components within the HRTFs primarily shaped by the outer ear and responsible for directional cues are selectively removed. This process is based on our novel auditory model, which builds on Gnther Theiles two-stage hearing framework. While Theile separates spatial and timbral processing via matrix encoding (M) and decoding (M?), our model introduces a customized mechanism that isolates the spectral components most essential for personal timbre recognition. The output of this process yields a Personalized Reference Target Response Curve (PR-TRC) optimized for each listener.
The fourth stage involves perceptual adjustment by applying a listener-specific weighting coefficient (0.00 to 1.00) to the PR-TRC. Unlike traditional preference-based tuning, this step is performed through structured listening tests that iteratively refine the coefficient to match the listeners natural timbral perception. The final target response curve is implemented as an FIR filter within the earphones digital signal processing (DSP) pipeline.
To evaluate the method, a two-part subjective study was conducted. In the first phase, participants compared audio processed with our personalized timbre curve to several industry-standard target curves using the Semantic Differential (SD) method. Statistically significant improvements (p < 0.01) were observed in many attributes. In the second phase, a six-week slow listening test was conducted, during which participants repeatedly evaluated multiple target curves. Only our method demonstrated sustained improvement in listener ratings over time, indicating strong perceptual validity beyond initial impressions.
This method has already been implemented in commercial true wireless stereo (TWS) earphones. Users consistently reported a more natural and emotionally engaging listening
experience with stereophonic music after undergoing this timbral personalization. These findings emphasize the importance of addressing timbre perception as a distinct
personalization domain, separate from spatial audio processing.
In summary, this work introduces and validates a novel method for earphone personalization focused on natural timbral reproduction of stereophonic content. By combining anatomical modeling, individualized auditory processing, and structured perceptual optimization not driven by preference but by perceptual accuracy our approach fills a critical gap in current headphone technology and offers a new direction for listener-centric sound design.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22947
(969KB)
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 Join the AES. If you need to check your member status, login to the Member Portal.
