An automated piecewise synthesis method for cetacean tonal sounds based on time-frequency spectrogram.

Area of Science:

• Bioacoustics
• Signal Processing
• Marine Biology

Background:

• Bionic signal waveform design is crucial for biological research and underwater acoustic applications.
• Existing methods struggle to accurately replicate complex cetacean sounds and modify signal time-frequency structures.
• Previous manual methods for synthesizing cetacean sounds were labor-intensive.

Purpose of the Study:

• To develop an automated method for synthesizing high-similarity bionic signals that match cetacean tonal sounds.
• To overcome the limitations of manual waveform construction and improve efficiency in bionic signal generation.
• To enable accurate imitation of diverse and complex cetacean vocalizations.

Main Methods:

• Automated recognition and extraction of fundamental and harmonic contours from time-frequency spectrograms of tonal sounds.
• Integration of four sub-power frequency modulation bionic signal models to match extracted sound contours.
• Synthesis of bionic signals by combining envelopes of fundamental frequency and harmonic contours.

Main Results:

• The automated method successfully extracts key acoustic features from cetacean sounds.
• The synthesized bionic signals demonstrate high similarity to original cetacean sounds.
• Pearson correlation coefficients (PCC) between true and synthesized sounds consistently exceeded 0.95.

Conclusions:

• The proposed automated piecewise synthesis method effectively imitates simple and complex cetacean tonal sounds with high fidelity.
• This advancement significantly enhances the capability for creating realistic bionic signals for research and applications.
• The method offers a more efficient and automated approach to bionic signal design compared to previous techniques.