Dolphin Sounds: Whistles, Clicks, and Burst Pulses in Cetacean Bioacoustics

Dolphin acoustic repertoires consist of three primary sound classes: frequency-modulated tonal whistles, broadband ultrasonic echolocation clicks, and rapid burst-pulsed packets. Each acoustic category possesses distinct time-frequency characteristics, sound production pathways, and behavioral contexts in marine environments.

Unlike terrestrial mammals, which produce vocalizations by forcing expired air across vocal folds in the larynx, odontocetes (toothed whales and dolphins) generate sounds through an internal pneumatic system located beneath the blowhole. Understanding dolphin sounds requires examining both their acoustic physics and their physiological sound generation mechanisms.

Acoustic Properties of the Three Primary Sound Classes

The acoustic space of delphinids spans from audible human hearing frequencies (a few kilohertz) up into extreme ultrasound (>130 kHz):

Acoustic Spectrum (Logarithmic Frequency Axis):
1 kHz              10 kHz             50 kHz             100 kHz            150 kHz
───┼──────────────────┼──────────────────┼──────────────────┼──────────────────┼───
   [ Tonal Whistles  ]──────────────────> [ Whistle Harmonics ]
   (4 – 24 kHz fundamental)
   
   [ Burst-Pulsed Sounds ]──────────────> [ Ultrasonic Packets ]
   (Repetition rates up to 2 kHz)
   
                      [ Broadband Echolocation Biosonar Clicks ]──────────────>
                      (Peak energy 40 – 130+ kHz; durations 40 – 100 μs)

1. Frequency-Modulated Tonal Whistles

Whistles are continuous, narrowband acoustic signals characterized by frequency modulation contours over time:

• Fundamental Frequency Range: Typically 4 kHz to 24 kHz in bottlenose dolphins (Tursiops truncatus) and Atlantic spotted dolphins (Stenella frontalis), though fundamental contours in delphinids can initiate as low as ~2 kHz or terminate above 28 kHz.
• Duration: Typically 0.2 to 2.5 seconds.
• Harmonic Structure: While early bioacoustic recordings with band-limited hydrophones suggested whistles were strictly low-frequency signals, research by Dr. Marc Lammers and Dr. Denise Herzing demonstrated that whistle harmonics regularly extend past 50 kHz to 100 kHz into ultrasonic frequencies.
• Behavioral Context: Whistles serve primarily as social contact calls, group coordination signals, and individual identity broadcasts. In bottlenose dolphins, individually stereotypic signature whistles can account for 50% to over 90% of whistles emitted during temporary isolation or brief capture-release health evaluations, whereas whistle repertoires in undisturbed free-swimming groups feature variable proportions of shared and non-signature whistles.

2. Broadband Echolocation Clicks

Clicks are transient, high-amplitude acoustic pulses used primarily for biosonar navigation and prey localization:

• Pulse Duration: Extremely brief in delphinids, lasting between 40 and 100 microseconds (μs). In contrast, narrow-band high-frequency (NBHF) odontocetes (such as harbor porpoises) emit longer pulses (~125–150 μs) without whistles.
• Peak Frequency: Concentrated in the ultrasonic spectrum between 40 kHz and 130+ kHz in broadband species.
• Directional Transmission: Clicks are focused into a narrow forward-facing acoustic beam by the lipid-filled melon on the dolphin's forehead, which acts as an acoustic impedance-matching lens.
• Inter-Click Interval (ICI): Dolphins adjust the time between clicks according to target distance, emitting the next click only after receiving the acoustic echo from the previous one (plus a brief neural processing latency of 15–45 ms).

3. Burst-Pulsed Sounds

Burst pulses occupy an intermediate acoustic territory between clicks and continuous tones:

• Acoustic Structure: They are click trains emitted at very high pulse repetition rates, sometimes exceeding 1,000 to 2,000 clicks per second (inter-click intervals below 5 milliseconds).
• Perception: Because the human ear cannot resolve individual acoustic events spaced closer than approximately 10 milliseconds, burst pulses sound like continuous buzzing, barking, squawking, or screaming tones.
• Behavioral Context: Burst pulses are strongly associated with high-arousal social behaviors, including agonistic interactions, disciplinary displays by adult dolphins toward calves, inter-pod dominance contests, and sexual signaling.

Cranial Sound Production and Biosonar Anatomy

The anatomical sound generation pathway in delphinids involves three coordinated structures:

1. Phonic Lips / Dorsal Bursae (PLDB): Located directly below the blowhole inside the upper nasal passages. High pressure forces air through muscular slit-like lips, causing the adjacent fatty dorsal bursae to vibrate and generate acoustic energy.
2. Melon (Acoustic Lens): A fatty, bulbous forehead organ composed of specialized low-density lipids that collimates high-frequency clicks into a directional biosonar beam, minimizing acoustic impedance mismatch between cranial tissue and surrounding seawater.
3. Mandibular Acoustic Window: Returning echoes are received not through external ear canals (which are vestigial and plugged with wax), but through thin, specialized "pan bones" in the lower jaw filled with acoustic intramandibular fat bodies that channel sound directly to the tympanoperiotic bone of the inner ear.

Methodological Implications for Computational Bioacoustics

The physical diversity of dolphin sounds presents specific challenges for machine learning and passive acoustic monitoring:

• High Sample Rates Required: To capture both tonal whistles (up to 24 kHz) and broadband echolocation clicks (up to 130 kHz), hydrophones and audio recording systems must sample at a minimum of 192 kHz to 384 kHz to prevent aliasing distortion.
• Multitask Neural Codecs: Modern neural audio codecs like SoundStream, utilized in the DolphinGemma research project, are trained to compress and tokenize both continuous tonal whistle contours and sharp ultrasonic click transients without requiring separate hand-crafted feature extractors.
• Scientific Limits: Classifying vocalizations into whistles, clicks, and burst pulses allows researchers to map acoustic density and behavioral state; it does not constitute language translation or semantic interpretation.