ABSTRAKSI: Audio watermarking merupakan suatu metode penyembunyian data tertentu ke dalam suatu file audio, tetapi tidak diketahui kehadirannya oleh indera manusia. Metode audio watermarking yang digunakan juga harus mampu menghadapi proses-proses pengolahan sinyal digital (resampling, compression,low pass filtering, dll) sampai pada tahap tertentu.
Dalam tugas akhir ini diimplementasikan metode audio watermarking yang memanfaatkan gabungan metode Direct Sequence Spread Spectrum dan Modified Discrete Cosine Transform dengan metode Complementary Filter Bank pada proses filtering. Audio yang digunakan sebagai file host berformat *.wav dengan data watermark berupa file binary image atau file text. Hasil dari sistem ini selanjutnya diuji secara objektif untuk mengetahui kualitas audio terwatermark dan data watermark. Sedangkan untuk mengetahui ketahanan data watermark terhadap pengolahan sinyal, dilakukan pengujian kompresi MP3, resampling, dan low pass filtering.
Dari pengujian didapatkan kualitas audio terwatermark yang dihasilkan sistem ini dipengaruhi oleh skala embedding, level filter, dan ukuran watermark. Kualitas audio terwatermark sangat baik didapatkan pada skala embedding = 4 dengan rata-rata nilai SNR sebesar 74,9 dB. Untuk menyeimbangkan aspek imperceptibility dan robustness, kombinasi optimal untuk sistem ini adalah level filter = 6 dan skala embedding = 3. Berdasarkan uji ketahanan, sistem ini cukup tahan terhadap kompresi MP3 dengan bitrate 128 kbps, Low Pass Filtering dengan cut-off 3 kHz serta resampling 22,05 kHz dan 11,025 kHz.
Kata Kunci : Audio watermarking, Direct Sequence Spread Spectrum, Modified Discrete Cosine Transform, Complementary Filter BankABSTRACT: Audio watermarking is a method of hiding data into an audio file, but its presence isn’t known by the human senses. The audio watermarking method must be able to deal with the processes of digital signal processing (resampling, compression, low pass filtering, etc.) until at a certain stage.
In this final report, it’s implemented an audio watermarking method based on combination of Direct Sequence Spread Spectrum and Modified Discrete Cosine Transform methods with Complementary Filter Bank for the filtering process. The audio which is used as the host file is *. wav format and the watermark is a binary image file or text file. The results from this system further tested to determine objectively the quality of the watermarked audio and the extracted watermark. Furthermore, to know the robustness of the watermark from digital processing, then the watermarked audio is tested by MP3 compression, resampling, and low pass filtering.
From the robustness testing, it’s obtained that the result of this system is influenced by embedding scale, filter levels, and the size of the watermark. The watermarked audio quality earned very good at embedding scale 4 with average SNR value of 74.9 dB. To balance the aspects of imperceptibility and robustness, the optimal combination for this system is filter level = 6 and embedding scale = 3. Based on the test of robustness, the system is robust enough to MP3 compression with bitrate 128 kbps, Low Pass Filtering with cut-off 3 kHz and resampling 22.05 kHz and 11.025 kHz.
Keyword: Audio Digital watermarking, Direct Sequence Spread Spectrum,Modified Discrete Cosine Transform, Complementary Filter Bank