Your session
Your session appears here
Describe what you are carrying — or pick an example — then generate to receive a named session with audio and guidance.
waveform appears after generation
R
Vers3Dynamics
Rhythma
Living modulation engine
Rhythma
Give your state a name.
Rhythma gives it a sound.
Describe what you're carrying, and Rhythma transforms it into a personalized listening experience through layered tone, movement, and reflection.
Layered audio
Session profiles
Reflective use
Inside the session
Rhythma keeps the interaction compact while exposing enough structure to make each generated session legible.
01
Input
A feeling or intention becomes the starting signal.
02
Profile
The analyzer selects a session title, tone center, pattern, brightness, and density.
03
Render
The engine blends a drone, breath pulse, and shimmer layer into one WAV session.
04
Listen
The final surface presents guidance, audio, waveform feedback, and a short reflection.
Demo surface
The core engine stays Python. The Space now serves a full-width FastAPI shell with a small JSON API.
app.py
FastAPI routes, session JSON, audio file serving, and the preserved legacy Gradio builder.
demo.html
Full-screen responsive interface with examples, controls, audio playback, and visual feedback.
rhythma_engine.py
Layered modulation renderer that creates the listening session behind each result.
Scientific Research Mode
New Mothers
Personalized sound research to discover gentle acoustic patterns that help individual babies calm down and settle faster.
How It Works & Safety Guidelines
Every baby is unique. This tool explores how personalized sounds help an individual infant settle: Baby's Cry State + Soothing Sound → Settling Response. Not a medical device. Audio is volume-protected with automatic rest breaks. Always keep room speaker volume low and supervise your baby.
Acoustic Input & Synthetic Signal
Upload an infant recording or generate a synthetic cry episode for offline testing.
Detection & Temporal Metrics
Ready for signal input
Total Length--
Cry Duration--
Pause Total--
Cry Rate--
Acoustic Feature Trajectory Extract
| Time (s) | F0 (Hz) | RMS Energy | Centroid (Hz) | Entropy | Harmonicity | Observed State |
|---|---|---|---|---|---|---|
| No audio analyzed yet. | ||||||
2D Acoustic State-Space Trajectory (PCA)
Continuous trajectory of cry dynamics projected from spectral, energy, and pitch features into 2D state space.
← Lower Arousal / Settled
State Vector Sequence: 0 frames
Higher Intensity / Vocalization →
Constrained Stimulus Designer
Synthesize reproducible resonant stimuli via Rhythma's modulation engine within safety limits.
Stimulus Verification & Playback
Stimulus engine ready
Stimulus ID--
Peak Amplitude--
Clearance Token--
Stimulus audio is stored in an ephemeral private cache and evicted automatically after 10 minutes.
Experimental Trial Runner
Trial Session Record
Awaiting trial run
Session ID--
Condition--
Baseline Cry Rate--
Post-Stim Cry Rate--
Settling Score--
Individualized Response Modeling
Fit linear and regularized models relating acoustic state & stimulus parameters to settling outcome.
Model ready for training
Training Trials--
MSE--
R² Score--
Settling Prediction Tool
Predict response outcome for a candidate stimulus given current infant state.
Predicted Settling Index--