Farhan

Posted on Nov 7 AI-assisted

PulseTones: Real-Time Bio-Acoustic Generative Soundscape with SmartSpectra SDK

#devchallenge #presagechallenge #ai #audio

Presage SmartSpectra SDK Challenge Submission

This is a submission for the Presage SmartSpectra SDK Challenge

What I Built

PulseTones is an interactive, contactless bio-acoustic experience that translates the hidden rhythms of the human body into a generative ambient music soundscape and bioluminescent visual installation.

Using an ordinary webcam and the Presage SmartSpectra SDK, PulseTones reads your heart rate, respiratory waveforms, HRV autonomic balance, and facial expressions in real-time through remote photoplethysmography (rPPG). Instead of displaying clinical medical charts, PulseTones turns your physiological state into an auditory breathing meditation:

Respiratory Swell: Inhaling opens an analog-style 24dB resonant lowpass filter, flooding the room with bright, shimmering harmonics; exhaling calms the soundscape into deep, grounding sub-bass roots.
Cardiovascular Rhythm: Your pulse rate drives a warm acoustic sub-bass pulse (~56Hz to 36Hz downward sweep) and synchronized bioluminescent shockwaves across the display.
Autonomic Shimmer: High HRV coherence introduces lush 9th and 11th modal chords with stereo spatial diffusion, easing cognitive tension.
Affect Chimes: Smiling triggers a cascade of crystalline pentatonic chime arpeggios that twinkle across the stereo field.

To make the app instantly accessible to anyone without requiring them to sit in front of a live camera or share personal video, PulseTones features a Photorealistic AI Benchmark Subject Demo Mode alongside full Live Camera Mode. The demo runs an authentic 512×512 biometric video loop overlaid with a real-time 3D facial landmark mesh, pulsing rPPG forehead and cheek ROI tracking boxes, and active sweep lasers.

Demo & Experience Showcase

Experience Highlights & Controls

Dual Elevated Oscilloscopes:
- ♥ Arterial Pulse (rPPG): High-cadence neon magenta oscilloscope rendering systolic peaks, dicrotic notches, and microvascular flow rate.
- ≋ Respiration Tidal Wave: Luminous cyan oscilloscope tracking thoracic displacement and resonant lowpass filter modulation.
Central Bioluminescent Sacred Mandala:
- Dynamic 60fps canvas engine expanding and contracting with respiration tidal volume while radiating shockwaves on each cardiac pulse.
Procedural Web Audio Engine:
- Switchable harmonic environments: Celestial Calm (C Minor 9), Deep Forest (D Pentatonic), and Ocean Mist (A Sus2).
Zero-Setup AI Simulation Mode:
- Pre-calibrated photorealistic AI benchmark subject built right into the app so judges and users can run npm start and immediately explore the bio-acoustic synthesis without needing webcam access or personal video recording.

Code

The complete source code is open source on GitHub:

👉 https://github.com/YTxFSGAMERz/pulse-tones

Quick Setup Instructions

# 1. Clone the repository
git clone https://github.com/YTxFSGAMERz/pulse-tones.git
cd pulse-tones

# 2. Install dependencies
npm install

# 3. Add your Presage API Key (or use in-app Settings dialog)
cp .env.example .env
# Edit .env with your SMARTSPECTRA_API_KEY

# 4. Start the application
npm start

How It Responds to Human Signals

PulseTones maps four continuous physiological signal streams from the Presage SmartSpectra SDK into procedural Web Audio API synthesis nodes and HTML5 Canvas visuals:

                      ┌───────────────────────────────────────────────┐
                      │          Presage SmartSpectra SDK             │
                      │       Remote Photoplethysmography (rPPG)      │
                      └───────┬──────────────┬─────────────┬──────────┘
                              │              │             │
                    Pulse Rate│   Respiration│          HRV│  Expression
                        (BPM) │       Waveform│      (RMSSD)│  (Smile / Serene)
                              ▼              ▼             ▼          ▼
        ┌─────────────────────┼──────────────┼─────────────┼──────────┐
        │                     │              │             │          │
        │  [Sub-Bass Thump]   │ [Biquad LPF] │ [Harmonics] │ [Chimes] │
        │  56Hz->36Hz Sine    │ 220Hz-2400Hz │ Modals 9/11 │ Pentatonic│
        │  Musical Clock      │ Resonance Q=4│ Stereo Ping │ Arpeggio │
        │                     │              │             │          │
        │               Web Audio API Procedural Synthesizer          │
        └─────────────────────────────────────────────────────────────┘

Pulse Rate (rPPG): The SDK detects microscopic capillary color oscillations across the forehead and cheek regions of interest (ROI). This BPM value synchronizes an internal high-resolution audio clock scheduling a dual-sine sub-bass pulse and triggers radial bioluminescent shockwaves on the canvas.
Respiratory Movement (Breathing Trace): Continuous tidal volume shifts dynamically sweep a 4-pole 24dB/octave resonant lowpass biquad filter. Inhaling lifts the cutoff from 220 Hz up to 2400 Hz (brightening the acoustic field); exhaling slowly rolls off the frequencies for parasympathetic relaxation.
Heart Rate Variability (HRV RMSSD): Measures inter-beat interval consistency. Higher autonomic coherence blends in suspended 9th and major 7th chord tones with ping-pong delay diffusion; lower coherence shifts the voicing to pure fifths to anchor and stabilize the listener.
Affect & Expression: When facial landmark valence indicates a smile, an expressive chime engine triggers high-register bell tones (FM synthesis with bell partials) panned across the stereo spectrum.

All SDK events (metrics, validationStatus, processingStatus, streamAvailable) are communicated from Electron's main process to the isolated renderer process via high-performance IPC channels with zero telemetry lag.

Prize Categories

Most Creative or Delightful Use

PulseTones reimagines biometric sensing not as a cold clinical diagnostic tool, but as an intimate, playful instrument of human expression and mindfulness.

Rather than confronting the user with anxiety-provoking medical meters and warning thresholds, PulseTones turns the body's subconscious rhythms into ambient music. You do not just see your vitals—you feel your heart setting the tempo of an ethereal chord progression, you hear your lungs modulating the warmth of the room, and a simple smile brightens the atmosphere with chimes.

By eliminating the need for awkward pulse oximeters, smartwatches, or chest straps, and by including a photorealistic AI benchmark subject for effortless zero-camera exploration, PulseTones offers a delightful, accessible glimpse into the future of ambient human-computer resonance.

Consent and Data Handling

Privacy and consent are fundamental architectural pillars of PulseTones:

Explicit Consent Gate: The application presents an unskippable Camera Consent Modal upon first launch. No camera devices are initialized or queried until the user explicitly clicks "I Understand & Consent".
Strictly Local In-Memory Processing: All video frames from the webcam are analyzed in volatile memory on-device by the native SmartSpectra SDK.
Zero Video or Biometric Storage: No camera frames, photos, recordings, or facial scans are ever written to disk, retained, or transmitted over the network.
No Facial Recognition or Tracking: The optical algorithm analyzes aggregate pixel reflectance for rPPG signal extraction; no biometric templates, facial recognition, or identity fingerprints are created.
Non-Medical Disclaimer: PulseTones is strictly an artistic generative soundscape and mindfulness aid. It is not a medical device, makes no diagnostic or therapeutic claims, and is not designed or intended for clinical decision-making.

Crafted with ❤️ for the DEV.to Community and the Presage SmartSpectra SDK Challenge.

DEV Community