Voice quality and understandability are both diminished by the telephone systems’ limited voice bandwidth. The innovative speech bandwidth enhancement approach is suggested in this paper. To deliver a superior broad-band voice signal, the methodology employs a spectral data masking approach. The restricted band speech signal conceals the missing speech's code-excited linear prediction (CELP) parameters. At the other end, the missing speech is retrieved to create a broad-band voice signal. The suggested approach is resilient to quantization and channel disturbances, according to theoretical and simulation evaluations. The obtained findings support the suggested method's superior reconstructed broad-band speech quality to conventional approaches.