Fig. 1. Illustrations of (a) holographic acoustic-signal authenticator and (b) holographic recording, visualization, and optical correlation. The acoustic target signal to be compared with corresponding reference signal is given as input to the optical machine containing the holography and correlator that can discriminate between acoustic signals. In holographic recording and visualization, a sequence of holograms is recorded for an acoustic signal and then the corresponding phase sequence is retrieved. Individual pixel values from the phase sequence can be appropriately arranged in two-dimensional (2D) format. Finally, optical correlation, which can be implemented by Fourier transform (FT) and inverse FT, provides correlation output.

Fig. 2. Proposed optical setup for implementing a holographic acoustic-signal authenticator. For voice recording, parallel phase-shifting digital holography is employed while optical correlation of voice signals is obtained by a joint transform correlator. NDF, neutral density filter; SF, spatial filter; CL, collimating lens; PBS, polarizing beam splitter; M, mirror; BS, beam splitter; SLM, spatial light modulator; L, lens; CCD, charge-coupled device camera; P, polarizer; QWP, quarter-wave plate; and HPC, high-speed polarization camera.

Fig. 3. Schematic diagrams of (a) phase image reconstruction from recorded spatially multiplexed hologram and (b) generating 2D voice image from reconstructed phase images. PPSDH, parallel phase-shifting digital holography. The 2D conversion follows a column-wise arrangement.

Fig. 4. Schematic diagram of fringe adjusted joint transform correlation. FT, Fourier transform; IFT, inverse FT; and JPS, joint power spectrum.

Fig. 5. Results of optical voice authentication. (a)–(c) are temporal phase profiles of experimentally recorded three voices corresponding to texts “Thank you,” “Hello,” and “Arigato,” respectively. (d)–(f) are corresponding 2D voice images.

Fig. 6. Results of JPS. (a)–(c) The obtained JPS corresponding to the text voices of “Thank you,” “Hello,” and “Arigato,” respectively, when the same voices are used with itself as a joint input; (d) the obtained JPS when the text voices “Thank you” and “Hello” are joint input; (e) the obtained JPS when the text voices “Thank you” and “Arigato” are joint input; and (f) the obtained JPS when the text voices “Hello” and “Arigato” are joint input.

Fig. 7. Results of optical voice authentication obtained by fringe-adjusted JTC. (a)–(c) 2D autocorrelation signals, (d)–(f) 2D cross-correlation signals, (g)–(i) 3D plots corresponding to (a)–(c), and (j)–(l) 3D plots corresponding to (d)–(f). In the case of autocorrelation, when the target and reference acoustic signals are the same, the peak signal can be seen, whereas in the case of cross-correlation, when they are different, the peak signal does not appear.

Fig. 8. Results of Gaussian and speckle noise attack. (a) 2D autocorrelation signals when Gaussian noise of 0.01 variance with mean of 0.2 is added to the voice image, (b) 2D autocorrelation signals when Gaussian noise of 0.01 variance with mean of 0 is added to the voice image, (c) simulated 3D plot corresponding to (a), (d) simulated 3D plot corresponding to (b), (e) 2D autocorrelation signals when speckle noise of variance 0.25 is added to the voice image, (f) 2D autocorrelation signals when speckle noise of variance 0 is added to the voice image, (g) simulated 3D plot corresponding to (e), and (h) simulated 3D plot corresponding to (f).

Fig. 9. Results of occlusion attack. (a) 2D autocorrelation signals when 5% of voice image is occluded, (b) simulated 3D plot corresponding to (a), (c) 2D autocorrelation signals when 25% of JPS is occluded, (d) 2D autocorrelation signals when 40% of JPS is occluded, (e) simulated 3D plot corresponding to (c), and (f) simulated 3D plot corresponding to (d).