<p>Head-related transfer functions (HRTFs) play a critical role in the construction of immersive audio. However, traditional HRTF measurement systems are typically limited to obtaining sparse HRTF data, which is insufficient to meet the high-precision spatial resolution requirements of virtual acoustics. Therefore, there is an urgent need to generate impulse responses (IR) for target directions through spatial interpolation techniques. This paper proposes an HRTF spatial interpolation method based on Enhanced super-resolution generative adversarial networks (ESRGAN). The method uses Residual-in-residual dense blocks (RRDB) as the basic unit of the network and introduces the concept of Relativistic average generative adversarial networks (RaGAN) to enable the discriminator to predict relative realism, thereby improving the realism and consistency of the generated results. In addition, wavelet transform is used to extract high-frequency and low-frequency information, enhancing the representation of details, especially in peak and valley regions. To further improve localization performance, source position information is integrated into the network to provide additional spatial cues. Numerical experimental results show that the proposed method outperforms SRGAN, barycentric interpolation, and HRTF selection methods in terms of log-spectral distortion (LSD) and perceptual evaluation metrics, demonstrating significant advantages in sparse interpolation tasks.</p>