Product Strategy & Development Document April 2026 — Confidential deepmade.ai
DeepMade is a generative soundscape platform engineered to enhance human performance through biometrically adaptive audio. It is built on the same neuroscience used in clinical music therapy [1][2], but positioned entirely differently.
Not as medicine, not as meditation, but as performance infrastructure for people who are moving, thinking, recovering, and living with intention.
The technology stack is ready. Apple HealthKit and Google Health Connect provide real-time biometric data access to over 95% of the global smartwatch install base. Web Audio API and native audio frameworks allow us to generate and deliver high-quality adaptive soundscapes to virtually every smartphone on the market. The infrastructure exists. The product does not.
The core proposition is simple: the right sound, generated in real time and calibrated to your physiological state, can help you focus deeper, sleep sounder, recover faster, and perform better. DeepMade is the product that will deliver that proposition at a quality level the market has never seen.
The underlying science is established. Our moat will exist not in the research itself, but in the integration of biometric data streams, the creation of new derived data from those inputs, and the real-time feedback loops that connect physiology to synthesis. This is an engineering and design problem, not a research problem.
Now is the best time to build this. Incumbents, whether wearable manufacturers, streaming platforms, or wellness apps, are integrating AI into pre-existing user experiences and business models. They are constrained by what they’ve already built. DeepMade has no legacy to protect. We are free to design for the terminal realities of the AI era: continuous sensing, real-time generation, and closed-loop systems that respond to humans rather than waiting for humans to respond to them.
The name carries a double meaning. Deep refers to the deep physiological states the product targets (deep focus, deep sleep, deep recovery) as well as the depth of neuroscience underpinning every synthesis decision. Made signals that this is engineered audio, not curated playlists or ambient wallpaper. Every soundscape is made in real time, made for you, made to work. The name positions DeepMade as precision-engineered performance infrastructure, distinct from the passive wellness aesthetic of competitors.
BRAND: DeepMade operates under the DeepMade.ai brand, a premium biometric music and wellness platform. The domains deepmade.ai and deepmade.com are registered and active.
Apple, Samsung, Garmin, and Whoop have spent a decade proving that consumers will wear biometric sensors every day. Over 500 million people now have real-time heart rate, HRV, and movement data streaming from their wrists. Nike and Peloton proved that performance-oriented consumers will pay premium prices for tools that make them better. The hardware layer is mature. The data is abundant. What’s missing is software that actually uses it.
The opportunity in one sentence: Half a billion people are generating continuous biometric data, and no product exists that transforms that data into real-time audio optimised for how they’re performing right now.
The functional audio category (Endel, Brain.fm, Focus@Will) has validated consumer willingness to pay for audio that claims to improve focus or sleep. But these products remain small: Endel has raised approximately $15M and reports 4 million downloads; Brain.fm operates as a modest subscription business. Neither has broken into mainstream scale, and neither has captured the performance-oriented consumer who wears a Whoop or trains with a Garmin. The category exists. The breakout product does not.
We believe Running and Recovery are the use cases that unlock breakout scale. Not because DeepMade is a running app, but because these contexts make the biometric feedback loop undeniable. When you’re running with audio that responds to your stride in real time, you feel the technology working. When recovery audio shifts as your heart rate drops, the effect is immediate and legible. These are proof points, not product boundaries.
Running is also where the competitive gap is widest. Endel’s passive ambient aesthetic doesn’t translate to athletic intensity. Spotify playlists don’t adapt. Nike Run Club gave up on music entirely. The runner who wears a Garmin and demands precision from their training has no audio product designed for them.
But the underlying technology, biometrically responsive generative synthesis, is state-agnostic. The same architecture that locks tempo to stride also modulates texture as HRV signals stress, or fades entrainment frequencies as sleep onset approaches. Running and Recovery demonstrate the technology. The platform serves every state.
If DeepMade succeeds, it becomes more than a consumer app. It becomes the intelligence layer that transforms biometric data into real-time audio optimisation.
Today, wearable manufacturers generate abundant physiological data but offer limited utility beyond dashboards and notifications. Apple Watch knows your heart rate, HRV, movement, and sleep patterns, but does nothing with that data in real time except display numbers. DeepMade represents a new category: biometric actuation, using continuous physiological data to do something useful in the moment, not just report it afterward.
The platform implications: - Integration layer: Garmin, Polar, Whoop, and Apple could integrate DeepMade’s synthesis engine directly, offering adaptive audio as a native feature - Hardware evolution: A future where wearables ship with DeepMade embedded, generating audio on-device rather than requiring a separate app - New sensor classes: If the technology proves out, DeepMade could develop specialised sensors optimised for audio generation. Not passive monitoring, but active performance enhancement
This positions DeepMade not as “a running app” but as foundational technology for a new relationship between biometric sensing and human performance.
KEY INSIGHT: The science is proven. The hardware is ubiquitous. What does not yet exist is a consumer product that closes the loop: taking real-time biometric data and using it to generate audio that responds to your physiology, designed for active performance contexts rather than passive desk or bed use cases.
DeepMade is grounded in three converging bodies of neuroscience research, each with substantial peer-reviewed evidence. Together they form a coherent architecture for how sound can modulate human physiological and cognitive states.
The human brain contains a deeply wired connection between auditory processing and motor output. When we hear a rhythmic pulse, neural circuits in the motor cortex, basal ganglia, and cerebellum synchronise automatically to that tempo. This is called the frequency-following response [3][4].
Brainwave entrainment is the synchronisation of neural oscillations to external auditory stimuli. Different frequency bands correspond to distinct cognitive and physiological states:
| Brainwave Band | Associated State |
|---|---|
| Delta (1–4 Hz) | Deep sleep, tissue repair, immune function |
| Theta (4–8 Hz) | Deep relaxation, creativity, REM sleep onset |
| Alpha (8–14 Hz) | Relaxed focus, stress reduction, learning readiness |
| Beta (14–30 Hz) | Active concentration, problem-solving, alertness |
| Gamma (30–100 Hz) | High cognitive load, memory consolidation, peak performance |
Binaural beats are created by presenting slightly different frequencies to each ear, causing the brain to perceive a third tone equal to the frequency difference [7]. This perceived tone drives entrainment toward the target brainwave state. For example, 200 Hz in the left ear and 208 Hz in the right creates a perceived 8 Hz binaural beat, nudging the brain toward alpha state [8].
Isochronic tones provide an alternative entrainment mechanism that functions without headphones, a significant practical advantage for athletic contexts.
A growing body of research specifically examines binaural and rhythmic audio in athletic contexts:
The science is real, but it is important to be precise about effect sizes and what mechanisms are strongest. This honesty strengthens the case for DeepMade by identifying where the largest opportunities lie.
| Intervention | Effect Size | Real-World Translation | Source |
|---|---|---|---|
| Generic music vs silence | g=0.31 (small-medium) | ~3-4% performance improvement | Terry et al. 2020 meta-analysis [5] |
| RPE reduction (low-moderate intensity) | ~10% | Same effort feels easier | Karageorghis 2012 review [6] |
| Synchronous tempo-matched audio | +7% efficiency | Better running economy | Bood et al. 2013 [14] |
| Motivational music (time to exhaustion) | +18-20% | Significantly longer endurance | PLOS One running study [14] |
| Binaural beats + music (combat sports) | η²p=0.29-0.33 (large) | Striking performance improvement | Frontiers 2025 [15] |
The critical insight: Generic music delivers modest benefits (3-4%). Synchronized, tempo-matched, biometrically-responsive audio delivers 7-18% improvements. That gap is DeepMade’s opportunity.
| Intervention | Effect | Source |
|---|---|---|
| Theta binaural beats post-exercise | Significant HF/LF HRV shift toward parasympathetic | Frontiers 2014 [16] |
| 0.25 Hz binaural beats | Shorter N2 and N3 latency (faster deep sleep onset) | Nature Scientific Reports 2024 [17] |
| Music intervention on HRV (meta-analysis) | Increased HFnu, decreased LFnu | Lin et al. 2024 [18] |
Brainwave entrainment is plausible but not consistently proven. Of 14 studies examining whether binaural beats reliably shift brain oscillations, 5 supported the entrainment hypothesis, 8 contradicted it, and 1 showed mixed results [19]. The frequency-following response exists in auditory cortex, but whether this spreads to global brainwave changes that drive behavioral effects is uncertain.
Our position: We include binaural beats as one layer of the synthesis architecture, but we do not stake the product on entrainment alone. The mechanisms we have strongest evidence for are:
No single mechanism delivers 20% improvement. But multiple mechanisms working together can compound:
| Mechanism | Evidence Strength | DeepMade Implementation |
|---|---|---|
| Auditory-motor sync | Strong | Tempo locked to real-time cadence |
| Arousal modulation | Strong | Phase-based harmonic progression |
| Distraction/attention capture | Strong | Continuous evolving texture |
| Respiratory entrainment | Moderate | Volume/texture swells at breath rate (recovery) |
| HRV biofeedback | Moderate | Real-time adaptation to recovery metrics |
| Brainwave entrainment | Weak-Moderate | Binaural/isochronic layer (hedge bet) |
| Leading tempo | Untested | Audio tempo gradually pulls user toward target cadence |
The pitch: 5% from tempo sync + 5% from arousal modulation + 5% from reduced perceived exertion = compound benefit that no single-mechanism product delivers. DeepMade stacks every proven mechanism into one integrated system.
The functional audio category has been developing for two decades. Several companies have built real products with real users. None has captured the full opportunity.
The most direct analogue. Endel has patented generative soundscape technology that adapts to time of day, weather, heart rate, and location. 4 million downloads. Deals with Warner Music for artist collaborations. Available on iOS, Android, Apple Watch, Alexa, and Apple TV. Raised approximately $15M to date.
Their weakness: fundamentally passive and ambient. The experience is designed for people sitting still, at a desk, in bed. Walking cadence is a peripheral input, not a core design principle. No serious athletic performance orientation. The audio quality, while functional, lacks the premium design intentionality that a performance-focused user expects.
Scale context: 4 million downloads over 6+ years represents modest traction. For comparison, Calm has 100M+ downloads; Headspace has 70M+. Endel has proven the concept but not achieved breakout scale.
The most scientifically serious focus and productivity player. Uses AI to generate audio with rhythmic neural phase locking designed to influence brainwave states. Strong peer-reviewed backing. Subscription business with solid retention.
Their weakness: purely static and functional. No biometric input. No adaptive real-time loop. No physical performance use case. The audio sounds algorithmic and functional, not desirable or emotionally resonant.
Scale context: Brain.fm operates as a profitable niche business but has not achieved mainstream consumer scale. They’ve proven willingness to pay but not mass-market appeal.
Toronto-based digital therapeutics company using emotion AI and binaural beats for dementia care and anxiety treatment. Clinically validated. Pursuing FDA approval and Medicare reimbursement. Featured on CNN. Backed by institutional health investors.
Their position: deliberately clinical, deliberately slow. Their regulatory path and target population are entirely different from DeepMade. However, their published clinical evidence directly validates the underlying technology DeepMade is building on.
| Product | Position | Limitation |
|---|---|---|
| Focus@Will | Curated music for attention | No generative engine, no biometrics, static playlists |
| myNoise | Customisable soundscape generator | Manual, no intelligence, no adaptation |
| Hemi-Sync | Original binaural beat programmes | 1970s product UX, no consumer app sophistication |
| Headspace / Calm | Meditation and sleep audio | Curated content, no generative or biometric layer |
Mapping the landscape against two axes (biometric responsiveness and performance orientation) reveals a clear gap:
THE GAP: Existing products either treat biometrics as ambient context (Endel) or ignore them entirely (Brain.fm). Existing products design for passive states (desk, bed) rather than active performance. No product combines real-time biometric feedback, generative synthesis, and a design sensibility built specifically for physical and cognitive performance. DeepMade owns that space.
Building conviction in DeepMade requires acknowledging the hard questions. Below are the primary risks we see and how the roadmap addresses each.
The concern: Users might not feel the audio responding to their physiology. If the adaptation is too subtle, the core value proposition disappears. If it’s too obvious, it feels gimmicky.
How we address it: PoC 2 (The Movement Loop) is specifically designed to test this. Running and rowing are ideal contexts because the feedback is immediately legible: tempo locks to stride, intensity shifts with heart rate. We measure both objective performance (pace, power) and subjective perception (RPE, enjoyment). If users cannot perceive the loop in these high-signal contexts, we know before building further.
The concern: The referenced studies demonstrate effects in controlled settings. Real-world performance during actual exercise, with wind noise, distractions, and variable effort, may not replicate lab results.
How we address it: We do not claim clinical outcomes. We claim perceived benefit and desirability. PoC 1 tests whether DeepMade sounds better than Endel. PoC 2 tests whether users prefer it during real activity. The science provides the architecture; user preference provides the validation.
The concern: Many runners avoid earbuds for safety (traffic awareness). Bone conduction has audio quality limits. This may narrow the addressable market for our flagship use case.
How we address it: We lead with rowing, cycling (indoor), and gym contexts where earbuds are standard. Running enters as a secondary hero use case with clear acknowledgment of the safety tradeoff. The same technology serves all movement contexts; we choose proof points strategically.
The concern: Apple, Spotify, or Endel could build the same thing with more resources. The moat is execution, not IP.
How we address it: Speed and focus are the moat. Incumbents are integrating AI into legacy products; we are building AI-native. Their incentives are to protect existing business models; ours is to capture the gap. First-mover advantage in a well-executed vertical beats second-mover with more resources in an unfocused portfolio.
The concern: Placebo effect is real. Users may report benefit because they expect benefit. This undermines the neuroscience credibility claim.
How we address it: We design PoCs with blind comparisons (PoC 1) and measurable proxies (PoC 2 pace/power data). We never claim medical outcomes. We position as performance infrastructure, not therapy. The standard we must meet is: users prefer DeepMade and return to it. That is a product success, not a clinical trial.
DeepMade organises the user experience around six distinct session states, each with a defined neurological target and synthesis profile:
| Session State | Neurological Target & Synthesis Approach |
|---|---|
| Moving | Auditory-motor sync at target cadence. Beta/gamma entrainment. Tempo tracks real-time HR and stride. Modes: rowing, running, walking, cycling. |
| Recovering | Alpha entrainment. Parasympathetic activation. Respiratory pacing at 4-6 breaths/minute. HRV-responsive intensity. |
| Sleeping | Full sleep cycle support from onset through wake (see Sleep Mode detail below). |
| Focusing | Beta entrainment with gamma micro-bursts. Sustained attention support. Cognitive load optimised. |
| Working | Alpha-beta boundary. Flow state induction. Adaptive to time of day. |
Sleep is a first-class feature, not an afterthought. The synthesis engine supports full overnight sessions that follow natural ultradian rhythms:
| Session Type | Duration | Phases | Target Use |
|---|---|---|---|
| Quick Drift | 15 min | Unwind > Descend > Drift | Fast sleep onset for good sleepers |
| Gentle Descent | 30 min | Settle > Release > Descend > Drift | Extended transition for anxious minds |
| Deep Preparation | 45 min | Arrive > Settle > Release > Descend > Drift | Full relaxation sequence |
| 6 Hour Sleep | 6 hr | Onset + 4 cycles (Light > Deep > REM) + Wake | Complete night with gentle wake |
| 7.5 Hour Sleep | 7.5 hr | Onset + 5 cycles + Wake | Full ultradian rhythm support |
Sleep cycle structure: Each 90-minute cycle progresses through light sleep (N1/N2), deep sleep (N3/SWS), and REM, with entrainment frequencies matched to each stage: - Onset: 8-10 Hz alpha descending to 4-5 Hz theta - Light sleep: 6-8 Hz theta/alpha boundary - Deep sleep: 1-2 Hz delta (lowest intensity, tissue repair and memory consolidation) - REM: 4-5 Hz theta (slightly elevated intensity for dream state support) - Wake: Gradual ascent through 8 Hz alpha to 11+ Hz beta
The audio adapts across the night: earlier cycles have longer deep sleep phases; later cycles have extended REM. Wake sequences begin 10-15 minutes before target wake time with graduated intensity increases.
Each session is generated in real time from five independent synthesis layers, mixed dynamically in response to biometric inputs and session state:
| Layer | Function |
|---|---|
| Layer 1 — Foundation | Low-frequency drone or tonal base. Sets the key and harmonic centre. Establishes the neurological anchor frequency. |
| Layer 2 — Rhythmic Engine | Tempo and pulse layer. Drives auditory-motor sync for movement. Adapts BPM to cadence input in real time. |
| Layer 3 — Entrainment | Binaural beat or isochronic tone layer, tuned to target brainwave state. Sits beneath conscious perception. |
| Layer 4 — Texture | Generative ambient and harmonic material. Provides musical interest and prevents listener fatigue. Evolves slowly over session. |
| Layer 5 — Dynamics | Real-time intensity and density modulation driven by biometric data. Heart rate, HRV, and movement intensity shape the mix. |
DeepMade is being built as a React Native application targeting iOS and Android simultaneously. The development sequence is:
The browser-based PoC demonstrates synthesis viability, but production deployment requires a native audio engine. The performance advantages are significant:
| Capability | Web Audio API | Native Engine (AVAudioEngine / Oboe) |
|---|---|---|
| Latency | 20-50ms typical | <10ms achievable |
| Background playback | Limited/blocked | Full support |
| System integration | None | Lock screen, CarPlay, Siri |
| Biometric sampling | Polling required | Push notifications available |
| CPU efficiency | JavaScript overhead | Metal/GPU acceleration |
Why latency matters for biometric responsiveness: At 180 BPM running cadence, each stride is 333ms. A 50ms audio latency means the beat arrives 15% late—perceptible and disruptive. At <10ms native latency, audio feels instantaneous and the motor synchronisation loop tightens.
Synthesis capabilities at native level: - Granular synthesis: Real-time granular processing with thousands of grains for organic texture evolution - Physical modeling: Computationally expensive but produces more natural-sounding timbres than additive synthesis - Spectral processing: FFT-based manipulation for psychoacoustic optimisation - Multi-channel rendering: Spatial audio, ambisonic encoding, binaural rendering
Background operation: Web Audio suspends when backgrounded on iOS. Native engines continue generating audio during screen-off, app-switch, and notification states—critical for a product designed for physical activity.
The native engine roadmap: 1. iOS: AVAudioEngine with AudioUnit processing chain 2. Android: Oboe library for low-latency audio across device fragmentation 3. Apple Watch: AVAudioEngine in watchOS for fully on-wrist generation 4. Cross-platform: C++ audio core with platform-specific bindings
Modern headphones (AirPods Pro, Sony WF-1000XM series, others) support head-tracked spatial audio. This opens psychoacoustic opportunities beyond stereo:
Positioning audio elements slightly ahead of the listener creates a subtle “pulling” sensation. Research in spatial audio perception shows that sounds originating from in front draw attention and suggest forward motion. For running and cycling, this can reinforce the sense of momentum.
Implementation approach: - Primary rhythmic elements positioned 15-30° forward of center - Binaural rendering via HRTF (Head-Related Transfer Function) - Head tracking maintains “forward” regardless of head position
As intensity increases, the soundscape can expand spatially: - Warmup phase: Audio concentrated in front, intimate staging - Push phase: Audio widens to 180° arc, enveloping presence - Recovery: Gradual narrowing back to centered, calming focus
This creates a physical sensation of “opening up” during high effort and “settling in” during recovery, reinforcing the physiological state through spatial perception.
For devices without head tracking, static binaural rendering still provides: - Accurate left/right binaural beat delivery (critical for entrainment layer) - Sense of depth and space superior to stereo panning - Forward/behind positioning for rhythmic elements
DeepMade requires an iconic audio signature—a 2-3 second sonic identity that introduces every session. This serves both brand recognition and neurological priming functions.
Research into audio logos reveals consistent patterns across the most successful examples:
| Brand | Duration | Notes | Key Characteristics |
|---|---|---|---|
| Intel | 3 sec | 5 | D-major, ascending, resolved |
| Netflix | 3 sec | 2 | “Ta-dum” with tension/release |
| Mastercard | 3 sec | 4 | Melodic, resolved on root |
| Nokia | 2.5 sec | 13 | Melodic phrase from Gran Vals |
Key design principles identified: 1. Duration: 2-3 seconds optimal—long enough for recognition, short enough for frequent use 2. Note count: 3-5 notes balances memorability with distinctiveness 3. Interval choice: Perfect fifth (3:2 ratio) universally perceived as stable and pleasing 4. Resolution strategy: Traditional logos resolve to root; incomplete patterns create anticipation (Zeigarnik effect)
The Zeigarnik effect describes how incomplete patterns are remembered better than completed ones. An audio signature that doesn’t quite resolve creates subconscious anticipation—the brain seeks completion. For a session introduction, this primes the listener for what follows rather than providing premature closure.
Applied to DeepMade: The signature ends on a note that implies continuation (major 2nd above root) rather than resolving to root. The session itself becomes the completion.
The implemented signature follows these specifications:
Why this design:
The “Deep” in DeepMade: The 55 Hz foundation is felt as much as heard. Sub-bass frequencies activate the vestibular system and create a visceral sense of grounding—you feel the product’s depth before the session begins.
Harmonic series = natural resonance: The harmonic series (1:2:3:4:5…) appears throughout nature—in resonating strings, vocal harmonics, and even architectural acoustics. The brain recognizes this pattern as organic and “right.” The blooming effect mimics how natural resonant systems build energy.
Filter sweep = awakening: The progressive opening of the filter mirrors the experience of coming into focus—starting narrow and expanding to full awareness. This creates a sense of “powering up” or “tuning in.”
Neurological priming: The sub-bass fundamental begins entraining the brain toward the session’s target state before the session formally begins. The 55 Hz fundamental is close to the gamma/high-beta boundary, subtly activating alertness pathways.
The Deep Resonance signature is designed to prime the mind for optimal session engagement:
| Mechanism | Effect | Supporting Research |
|---|---|---|
| Sub-bass activation | Vestibular system engagement creates physical grounding sensation | Vestibular-auditory interactions (Todd & Cody, 2000) |
| Harmonic series recognition | Brain processes natural harmonic relationships with less cognitive load | Auditory scene analysis (Bregman, 1990) |
| Ascending energy | Rising harmonic bloom elevates arousal state in preparation for activity | Arousal modulation via tempo/pitch (Husain et al., 2002) |
| Frequency following prep | Sub-bass fundamental begins entrainment process before binaural beats engage | Neural entrainment literature (Chaieb et al., 2015) |
| Anticipatory attention | The 0.5s confirmation tone creates a “ready” signal, focusing attention | Temporal expectation and attention (Nobre & van Ede, 2018) |
The 1.5-second buffer between signature end and session audio start allows the reverb tail to decay naturally while the brain transitions from “signature recognition” to “session engagement.” This prevents the jarring experience of overlapping audio and gives the nervous system time to shift modes.
The session outro provides bookend symmetry with the intro signature, creating a clear sonic boundary that signals session completion:
Why this design:
Inverse bloom = completion: Where the intro blooms upward (awakening), the outro converges downward (settling). This creates mirror symmetry—the brain recognizes the session as a complete unit.
Bright resolution: Unlike the intro which ends on an unresolved note (Zeigarnik anticipation), the outro resolves to a bright major chord. The session has been completed; the brain receives closure.
Crossfade integration: Session audio fades to 10% volume over 0.5 seconds while the outro begins, preventing jarring silence before the signature plays.
Celebratory tone: The high sparkle frequencies create a subtle “achievement” feeling—you’ve completed something. This positive association encourages session completion and return usage.
The intro/outro pair creates a clear psychological container: Deep Resonance announces “we’re beginning,” and Complete announces “we’re done.” This ritualistic framing enhances both the perceived value of each session and the likelihood of habitual return.
The signature is currently implemented in the browser PoC and plays at session start. Future refinements may include: - Mode-specific variations (deeper/slower for sleep, brighter/faster for activity) - Spatial rendering (signature expands from center to surround field) - Haptic pairing for Apple Watch (vibration pattern synchronized to audio) - Abbreviated 2-note version for notifications
Mastercard’s sonic branding development reportedly took 2 years and involved testing across cultures, contexts, and formats. The investment reflects audio identity’s strategic value: every session start, every notification, every app launch reinforces brand recognition through a channel competitors cannot easily replicate.
DeepMade’s signature is designed to work across: - Session introduction (full signature) - Notifications (abbreviated 2-note version) - Phase transitions (textural echo of signature intervals) - Marketing content (recognizable audio watermark)
The synthesis engine is designed with efficacy measurement built in. Real-time audio parameters are displayed to users, and data collection requirements are defined for each mode:
| Category | Metrics | Availability |
|---|---|---|
| Biometric Inputs | HR, HRV, Movement, Breathing Rate, SpO2, Skin Temp, Sleep Stage | HR/HRV/Movement available now via HealthKit; Sleep Stage requires Watch integration |
| Outcome Metrics | Time to sleep onset, Wake episodes, Time in deep sleep, Time in REM, Sleep efficiency %, Morning HRV, Morning resting HR, Subjective quality rating | Derivable from available biometrics + user input |
| Learning Signals | Optimal onset Hz curve, Personal cycle length, Deep sleep timing preferences, Wake window sensitivity, Volume threshold, Chronotype adjustment | Requires longitudinal data collection |
| Category | Metrics | Availability |
|---|---|---|
| Biometric Inputs | HR, HRV, Cadence/Pace, Movement, Power/Watts, GPS | HR/HRV/Cadence/Movement available now; Power requires compatible sensors |
| Outcome Metrics | Avg HR per phase, Cadence lock %, Perceived effort (RPE), Performance vs target, Post-session HRV | Derivable from available biometrics + user input |
| Learning Signals | Optimal tempo lead rate, Intensity response curve, Phase duration preferences, Warmup length optimization, Peak HR correlation | Requires longitudinal data collection |
| Category | Metrics |
|---|---|
| Biometric Inputs | HR, HRV, Breathing Rate, Skin Temp |
| Outcome Metrics | HRV recovery rate, HR recovery curve, Time to baseline, Subjective relaxation rating |
| Learning Signals | Optimal breath rate, Best entrainment Hz, Session length preference |
Data availability key: - Available now via HealthKit/Health Connect - Planned integration (native app) - Critical for feature (blocks functionality without it)
The PoC displays a real-time “Current Sound Profile” panel showing entrainment Hz, brainwave band, filter frequency, waveform type, intensity, and tempo/breath rate. A “Data Collection for Efficacy” panel shows context-aware data requirements that change based on mode (sleep vs activity).
DeepMade is currently in active early development. The following has been completed or is in progress:
Every existing product in this space is designed for people sitting still. Endel is for your desk. Brain.fm is for your desk. DeepMade is the first generative audio product designed specifically around physical and cognitive performance states: training, deep work, recovery, sleep. The session state architecture reflects how humans actually move through their day.
Endel accepts heart rate as an input but treats it as ambient context. DeepMade uses biometric data as the primary driver of real-time synthesis. When your heart rate climbs during exercise, the rhythmic layer accelerates and the intensity layer responds. When recovery begins, the synthesis shifts. The music is a direct reflection of your physiology, not a playlist that happens to know your heart rate.
The functional audio category suffers from uniformly poor product aesthetics. Brain.fm sounds like a tool. Endel sounds pleasant but generic. DeepMade is being built with the same design seriousness as the best wearable technology. Every sonic decision is intentional, every interaction is considered. The product should feel as desirable as Whoop or Oura, not as functional as a white noise machine.
Most consumer products in this space use binaural beats as a marketing claim, a buzzword on the App Store listing, rather than a core design principle. DeepMade’s entrainment layer is engineered to specific neurological targets for each session state, using the same frequency protocols validated in clinical settings. The difference between an 8 Hz alpha target and a 20 Hz beta target is not cosmetic. It is the difference between a relaxation session and a focus session, built into the synthesis itself.
LUCID’s clinical path targets a small, high-friction market. Existing functional audio apps ignore the half-billion people wearing biometric sensors daily. DeepMade targets healthy, performance-oriented consumers: athletes, knowledge workers, students, anyone seeking an edge in how they perform and recover. The total addressable market spans wellness, fitness, productivity, and sleep, each a multi-billion dollar category. And critically, the biometric hardware is already on their wrists.
Traditional music is popular for exercise and focus, but any performance benefit is accidental. A song might happen to have the right BPM for your running cadence. A familiar track might trigger memory associations that improve mood. These effects are real but unintentional, inconsistent, and impossible to optimise.
More critically, traditional music comes with structural constraints that make it unsuitable for adaptive performance audio:
Licensing is prohibitively complex. Every song involves multiple rights holders (composition, recording, publishing), each requiring separate negotiation. Streaming services spend billions annually on licensing. For a startup, the legal and financial overhead of music licensing is a distraction from the core technology problem.
Adaptation doesn’t scale. Some products layer secondary audio (binaural beats, tempo adjustments) over licensed recordings. This approach is fundamentally limited. You cannot meaningfully adapt a finished recording to real-time biometric input without degrading it. The song was mixed for aesthetic purposes, not physiological ones. Layering entrainment frequencies on top creates interference, not integration.
The artist is the bottleneck. Every licensed track requires someone to have written, recorded, and cleared it. DeepMade’s generative engine produces unlimited original material tuned to specific physiological targets. There is no catalogue to exhaust, no back-catalogue licensing deal to negotiate, no dependence on external creative supply chains.
An intriguing strategic question: can DeepMade participate in traditional music distribution as a copyright holder rather than a licensee?
Generative audio that DeepMade creates is original work. Under current copyright frameworks, the entity that directs the creative process (selects parameters, curates outputs, makes aesthetic decisions) holds rights to the resulting material. DeepMade could generate curated “albums” or “playlists” of standalone soundscapes, register them as original works, and distribute through Spotify, Apple Music, and other streaming platforms.
This inverts the typical music tech relationship. Instead of paying for content, DeepMade earns from it. Instead of negotiating with labels, DeepMade becomes a micro-label. The brand built through the performance app creates demand; the streaming presence generates passive revenue and additional discovery.
This is not core to the immediate product roadmap, but it represents a potential revenue diversification and brand extension once the primary synthesis engine is proven and the DeepMade brand has recognition among performance-oriented listeners.
Most apps have visual logos. DeepMade has a sonic signature—a 2.5-second audio mark that introduces every session. This serves dual purposes:
Neurological priming: The signature prepares the brain for the session ahead. The ascending contour and deliberate non-resolution (ending on major 2nd rather than root) creates anticipation via the Zeigarnik effect. The user’s brain seeks completion—which the session provides.
Brand recognition: Intel, Netflix, and Mastercard have proven that audio logos create instant recognition across contexts. DeepMade’s signature is designed to be as recognizable on a podcast ad as at session start. Every use reinforces the brand in a channel competitors cannot easily replicate.
The signature is not an afterthought. It is engineered with the same psychoacoustic intentionality as the synthesis engine itself: specific frequencies, layered timbres, calculated intervals. This attention to sonic identity mirrors DeepMade’s broader commitment to audio that is made with purpose, not generated by accident.
POSITIONING: Apple proved people will wear sensors. Whoop proved athletes will pay for performance data. Endel and Brain.fm proved people will pay for functional audio. DeepMade is what happens when you connect these pieces: real-time biometric data driving premium generative audio, designed for the way performance-oriented humans actually live.
The PoC sequence is designed to prove three things in order: the synthesis quality exceeds existing alternatives; the biometric loop produces a measurable effect on human performance; and users return habitually. Each stage builds evidence for the next and maps to a fundraising or partnership moment.
Can we make something that sounds meaningfully better than Endel?
The Tone.js browser PoC currently in development. Build two 20-minute movement sessions (one at 160 BPM target, one at 140 BPM) and conduct blind listening tests against Endel equivalents with 5–10 participants.
This is not yet a neuroscience test. It is a quality and desirability test. The question is whether DeepMade sounds premium, intentional, and different, not just functional.
SUCCESS METRIC: 7 out of 10 listeners prefer DeepMade over Endel in a blind listening test.
Does biometric input change the experience in a perceptible and measurable way?
The flagship PoC. Integrate Apple Watch or phone accelerometer for cadence input, plus BLE heart rate from any compatible device. The synthesis engine shifts BPM and intensity layer in response. Run 10 participants over the same activity three times: audio off, Endel, and DeepMade with live biometric input.
Measure performance metrics, Rate of Perceived Exertion (RPE), and subjective enjoyment. Physical activity is the ideal PoC context because the biometric feedback loop is immediately legible. You can feel the audio responding to your movement.
SUCCESS METRIC: Average RPE lower with DeepMade than Endel at equivalent effort, or performance higher at equivalent RPE. Even directional results from 10 participants constitute significant evidence.
Does the same engine produce genuinely distinct experiences across all five states?
Once movement is validated, build one session each for Sleep, Focus, Relax, and Study using the same synthesis architecture but different parameter profiles. This is not a full product build. It is proof that the engine generalises without the experiences bleeding into each other.
SUCCESS METRIC: 10 users can correctly identify which state a session is designed for without being told.
Will people actually come back?
Wrap the movement PoC in a minimal React Native shell. Release to 50 users for four weeks with open access. Track session completion rate and weekly return rate. This is the hardest PoC and the one that determines whether DeepMade is a product or a demo.
SUCCESS METRIC: 40% or more of users complete at least one session in week 3 of the trial.
| Phase | Milestones |
|---|---|
| Now — Q2 2026 | Tone.js browser PoC. Sound quality validation. Informal blind listening tests. Iterate on synthesis until audio quality bar is confirmed. |
| Q3 2026 | Movement PoC with BLE heart rate integration. 10-person informal trial. RPE and performance data collected. Synthesis parameters refined based on real feedback. |
| Q4 2026 | Generalise to Sleep and Focus sessions. React Native shell. Closed beta of 50 users. Retention data collected. Fundraising or partnership conversation begins. |
| H1 2027 | Full five-state product. Apple Watch native integration. Proper biometric onboarding. Public launch with movement as the hero use case. |
The music tech funding landscape in 2024–2025 has shifted toward companies that work with the industry rather than disrupting it, and toward products with proven revenue models. DeepMade sits at the intersection of two well-funded categories (functional audio and health/wellness tech) while occupying a position neither has claimed.
DeepMade does not require music licensing, does not compete with labels, and does not generate content that displaces human artists. It generates original synthesis in real time from a parametric engine, a fundamentally different model from Suno or Udio. This sidesteps the entire rights management problem that has consumed the generative music category.
The moats are: the synthesis architecture and its biometric integration layer; the clinical science underpinning each session state; and, once established, the community and habitual use patterns of a performance-oriented user base.
THE PITCH: Half a billion people wear biometric sensors. Whoop proved they’ll pay for performance insights. Brain.fm proved they’ll pay for functional audio. Nobody has connected these: using real-time biometric data to drive audio that actually responds to your physiology, designed for active performance rather than passive listening. DeepMade is that product.
The functional audio space has established patent activity. Understanding what is already protected—and what is not—is essential for DeepMade’s IP strategy.
Endel Sound GmbH has filed at least 8 patents. Their primary patent application (US20200142371, filed 2019) covers:
Key observation: Endel’s patent focuses on passive contextual adaptation—the environment responds to ambient conditions and physiological state. It does not cover active performance mechanisms or the specific innovations DeepMade is developing for athletic and cognitive performance contexts.
Binaural beat technology itself is well-established prior art and cannot be patented. Key foundational patents include:
What can be patented: novel methods of generating, combining, or applying binaural beats in specific ways tied to biometric feedback or performance contexts.
Brain.fm has patents covering their specific neural phase-locking approach to focus audio. Their IP is oriented toward desk-based productivity rather than physical performance.
The following innovations appear to be novel and not covered by existing patents. These represent DeepMade’s potential IP position:
What it is: Audio tempo that gradually pulls the user toward a target cadence rather than simply matching current cadence.
Why novel: Existing adaptive audio matches user state. Leading tempo actively influences user state by incrementally shifting BPM (e.g., 2-3 BPM per minute) to guide the user toward optimal performance cadence.
Claims territory: Method for modifying human movement cadence through progressive tempo adjustment in real-time generated audio, using biometric feedback to calibrate pull rate.
What it is: A synthesis engine that simultaneously applies multiple entrainment and performance mechanisms, weighted dynamically based on real-time biometric input.
Why novel: Existing products apply single mechanisms (binaural beats OR tempo matching OR ambient sound). DeepMade’s architecture layers 6+ mechanisms (auditory-motor sync, arousal modulation, brainwave entrainment, respiratory pacing, attentional capture, texture evolution) with real-time weighting based on physiological state.
Claims territory: System for generating performance-optimized audio through concurrent application of multiple neurological entrainment mechanisms, with dynamic weighting determined by continuous biometric feedback.
What it is: Soundscape spatial field that expands and contracts based on exercise intensity.
Why novel: Spatial audio is typically static or head-tracked for realism. This innovation uses spatial positioning as a performance mechanism—widening the sound field during high effort to create sensation of “opening up” and narrowing during recovery for calming focus.
Claims territory: Method for modulating spatial audio field geometry based on physiological effort metrics to enhance perceived performance state.
What it is: Positioning primary rhythmic elements 15-30° ahead of the listener to create forward momentum sensation.
Why novel: Spatial audio typically aims for neutral or realistic positioning. This deliberately positions sound ahead to psychoacoustically reinforce forward movement during running, cycling, and rowing.
Claims territory: Method for enhancing perceived forward momentum through anterior spatial positioning of rhythmic audio elements during locomotion activities.
What it is: Rhythmic audio patterns tied to specific phases of movement cycles (rowing stroke: catch, pull, release; running gait: footstrike, flight).
Why novel: Existing tempo-matched audio provides steady beats at target BPM. This innovation creates rhythmic textures that align with the phases within each movement cycle, reinforcing motor patterns at the micro-timing level.
Claims territory: System for generating audio synchronized to discrete phases within human movement cycles, using motion sensor data to identify and align with individual movement components.
What it is: Harmonic content that shifts from consonant (warmup) to dissonant (push) to resolved (cooldown) based on workout phase.
Why novel: Existing adaptive audio adjusts tempo and intensity. This innovation uses harmonic tension as a performance lever—dissonance during peak effort creates productive tension; resolution during cooldown facilitates parasympathetic activation.
Claims territory: Method for modulating harmonic consonance in generated audio based on exercise phase to optimize physiological state transitions.
What it is: Session introduction audio signature designed with deliberate non-resolution to create neurological anticipation.
Why novel: Audio logos typically resolve to tonic for completion and satisfaction. This innovation applies the Zeigarnik effect (incomplete patterns create anticipation) to session introductions, priming the brain for the session that follows.
Claims territory: Method for neurological session priming through audio signature with intentional harmonic non-resolution.
File provisional patent applications for the three highest-value innovations:
Cost: ~$1,000 total ($320 filing fee each) Protection: 12-month “patent pending” status; establishes priority date
Commission professional prior art searches for provisional applications before converting to non-provisional. This identifies potential conflicts and strengthens final claims.
Cost: $1,500-3,000 per application
Convert strongest provisionals to full patent applications. Timing aligned with product launch and potential fundraising.
Cost: $15,000-25,000 per application (including attorney fees) Timeline: 2-4 years to grant
Trade secrets: The specific parameter values, weighting algorithms, and synthesis recipes may be better protected as trade secrets than patents. Patents require public disclosure; trade secrets do not.
Trademarks: “DeepMade” and the sonic signature should be registered as trademarks. The audio logo can be registered as a sound mark (see Intel, Netflix precedents).
Defensive publication: For innovations we choose not to patent, defensive publication establishes prior art and prevents competitors from patenting the same concepts.
DeepMade’s path to market validation and scientific credibility runs through strategic partnerships with research institutions and technology providers. Two partnership categories are prioritized: clinical validation partners and neural interface integration.
Advisor: Dr. Jenny C. Chang, MB.ChirB., MD, MCHM — Ernest Cockrell, Jr. Presidential Distinguished Chair; President and CEO, Houston Methodist Academic Institute; Executive Vice President and Chief Academic Officer, Houston Methodist.
Houston Methodist’s mission—“driving science with purpose toward clinical solutions” and “rigorously validated innovation”—aligns directly with DeepMade’s need to move beyond “feels like it works” to clinically measured outcomes.
Houston Methodist provides three critical capabilities:
| Study | Measures | DeepMade Benefit |
|---|---|---|
| Sleep Lab Validation | Polysomnography, sleep onset latency, time in deep/REM | Validate entrainment frequencies actually shift sleep architecture |
| Recovery Protocol RCT | HRV recovery curves, cortisol, lactate clearance | Prove parasympathetic activation claims with biomarkers |
| Athletic Performance Trial | VO2max efficiency, RPE at fixed output, time to exhaustion | Quantify the 7-18% performance improvement hypothesis |
| Cardiac Rehabilitation | Post-procedure recovery metrics, anxiety reduction | Opens medical/clinical revenue stream |
Houston Methodist’s existing relationships with elite athletes provide:
| Before Houston Methodist | After Houston Methodist |
|---|---|
| “DeepMade uses neuroscience principles validated in other contexts” | “DeepMade is clinically validated at Houston Methodist Academic Institute” |
This is the credibility gap between Endel’s positioning and a defensible performance product.
Phase 1 — Pilot Study (Q3-Q4 2026): 30-person RCT on recovery audio vs control, measuring HRV recovery rate post-exercise. Low cost, high signal, fast execution.
Phase 2 — Sleep Architecture Study (2027): Proper polysomnography study validating sleep mode entrainment frequencies. Publishable in peer-reviewed journals.
Phase 3 — Elite Athlete Program: Partner with Houston Methodist sports medicine on professional athlete recovery protocols. Generates testimonials, case studies, and longitudinal data.
Funding pathway: NIH, DoD (military performance/recovery applications), or sports federation grants. Houston Methodist brings established grant infrastructure and investigator credentials.
| Opportunity | Description |
|---|---|
| B2B Sports Licensing | License to professional teams, collegiate athletics via Houston Methodist network |
| Clinical Deployment | Cardiac rehabilitation, post-surgical recovery, anxiety management protocols |
| Enterprise Wellness | Employer wellness programs with clinical backing for ROI documentation |
| Digital Therapeutics | Long-term FDA pathway for specific clinical claims (similar to LUCID’s regulatory approach) |
The long-term vision for DeepMade includes integration with brain-computer interfaces (BCIs) for direct neural feedback. This section outlines the strategic approach across near-term and long-term horizons.
Current DeepMade architecture infers brain state from proxy signals (HRV, movement, time). Direct neural measurement would enable:
| Capability | Application |
|---|---|
| Direct neural readout | Measure actual delta/theta/alpha/beta power instead of estimating from HRV |
| Closed-loop precision | Audio adapts to measured brainwave state in real-time |
| Entrainment validation | Prove definitively whether binaural beats shift neural oscillations |
| Sub-second latency | Neural feedback → audio adjustment in milliseconds |
| Personalization | Adapt to individual neural response patterns |
This addresses the contested science directly: the strategy document notes 8 of 14 studies contradicted the brainwave entrainment hypothesis. With direct neural measurement, DeepMade could know in real-time whether entrainment is working and adapt until it does.
What Neuralink provides: High-resolution neural recording via implantable BCI; eventual consumer pathway for cognitive enhancement applications.
Current limitations: - Focus remains on medical applications (paralysis, neurological conditions) - FDA approval pathway for implantables is slow and complex - Consumer BCIs are 5-10+ years from mainstream adoption - Brain surgery for better sleep audio is a difficult value proposition
The long-term play: Position DeepMade as the audio intelligence layer for neural interfaces. When BCIs reach the enhancement market, DeepMade should be the obvious audio integration.
Potential pitch to Neuralink:
“Audio is the lowest-friction, highest-impact modality for neural modulation. DeepMade has built the adaptive audio engine; Neuralink has the neural interface. A research collaboration—using Neuralink’s neural recording to validate and optimize entrainment algorithms—generates publishable science and positions the integration for when BCIs reach the enhancement market.”
More immediately actionable partnerships exist with consumer and research-grade EEG devices:
| Company | Device | Integration Opportunity |
|---|---|---|
| Muse | Consumer EEG headband | Sleep/meditation validation with direct brainwave measurement; large existing user base |
| Emotiv | Research-grade EEG | Lab studies on entrainment efficacy; scientific publication pathway |
| Kernel | Flow helmet (fNIRS + EEG) | High-end neural imaging for premium research validation |
| OpenBCI | Open-source EEG | Proof-of-concept integration; developer community engagement |
Recommended approach: Partner with Muse or Emotiv in 2026-2027 to validate entrainment with direct neural measurement. This generates evidence that: 1. Validates DeepMade’s entrainment claims with neural data 2. Differentiates from competitors who lack neural validation 3. Positions for future Neuralink/BCI integration when those platforms mature
DeepMade should be built as the audio intelligence layer that works across interface evolution:
| Era | Interface | DeepMade Role |
|---|---|---|
| Now (2026) | Wearables (Apple Watch, Whoop, Garmin) | Biometric-adaptive audio from HRV, movement, heart rate |
| Near-term (2027-2028) | Consumer EEG (Muse, Emotiv) | Neural-validated entrainment; closed-loop alpha/theta targeting |
| Medium-term (2029-2032) | Advanced non-invasive (Kernel, next-gen EEG) | High-resolution neural feedback; personalized entrainment curves |
| Long-term (2033+) | Implantable BCIs (Neuralink, competitors) | Direct neural integration; verified closed-loop brain-audio system |
The synthesis engine architecture remains constant; the input signals evolve. Each interface generation provides higher-fidelity neural data, enabling more precise audio adaptation.
| Timeline | Partnership | Objective |
|---|---|---|
| Q3 2026 | Houston Methodist | Clinical validation, professional sports access |
| Q4 2026 | Muse or Emotiv | Non-invasive neural validation pilot |
| 2027 | Publish research | Peer-reviewed evidence from both partnerships |
| 2028+ | Neuralink exploratory | Position for BCI integration when consumer pathway emerges |
PARTNERSHIP THESIS: Houston Methodist validates that DeepMade works physiologically. Muse/Emotiv validates that it works neurologically. These evidence bases position DeepMade as the credible audio layer when Neuralink and competitors bring BCIs to the enhancement market. The partnerships are sequenced to match technology readiness and DeepMade’s stage of development.
[1] Thaut, M.H. (2015). “The discovery of human auditory-motor entrainment and its role in the development of neurologic music therapy.” Progress in Brain Research, 217, 253-266. doi:10.1016/bs.pbr.2014.11.030
[2] Bradt, J., Dileo, C., Magill, L., & Teague, A. (2016). “Music interventions for improving psychological and physical outcomes in cancer patients.” Cochrane Database of Systematic Reviews. doi:10.1002/14651858.CD006911.pub3
[3] Ross, B., & Bhattacharyya, A. (2021). “Auditory-motor coupling in the human brain.” Trends in Neurosciences, 44(7), 534-544. doi:10.1016/j.tins.2021.03.004
[4] Grahn, J.A., & Brett, M. (2007). “Rhythm and beat perception in motor areas of the brain.” Journal of Cognitive Neuroscience, 19(5), 893-906. doi:10.1162/jocn.2007.19.5.893
[5] Terry, P.C., Karageorghis, C.I., Curran, M.L., Martin, O.V., & Parsons-Smith, R.L. (2020). “Effects of music in exercise and sport: A meta-analytic review.” Psychological Bulletin, 146(2), 91-117. doi:10.1037/bul0000216
[6] Karageorghis, C.I., & Priest, D.L. (2012). “Music in the exercise domain: A review and synthesis.” International Review of Sport and Exercise Psychology, 5(1), 44-66. doi:10.1080/1750984X.2011.631027
[7] Oster, G. (1973). “Auditory beats in the brain.” Scientific American, 229(4), 94-102. doi:10.1038/scientificamerican1073-94
[8] Chaieb, L., Wilpert, E.C., Reber, T.P., & Fell, J. (2015). “Auditory beat stimulation and its effects on cognition and mood states.” Frontiers in Psychiatry, 6:70. doi:10.3389/fpsyt.2015.00070
[9] Lane, J.D., Kasian, S.J., Owens, J.E., & Marsh, G.R. (1998). “Binaural auditory beats affect vigilance performance and mood.” Physiology & Behavior, 63(2), 249-252. doi:10.1016/S0031-9384(97)00436-8
[10] Shekar, D., Reddy, K.J., & Gourishankar, A. (2018). “Effect of alpha-frequency binaural beats on heart rate variability during recovery from exercise.” Journal of Clinical and Diagnostic Research, 12(6), CC01-CC04. doi:10.7860/JCDR/2018/34702.11612
[11] Reedijk, S.A., Bolders, A., & Hommel, B. (2013). “The impact of binaural beats on creativity.” Frontiers in Human Neuroscience, 7:786. doi:10.3389/fnhum.2013.00786
[12] Russo, F.A., Vempala, N.N., & Sandstrom, G.M. (2013). “The effect of music with lyrics on cognitive performance.” LUCID Audio Lab, Toronto Metropolitan University. Research in progress.
[13] Thaut, M.H., McIntosh, G.C., & Hoemberg, V. (2014). “Neurobiological foundations of neurologic music therapy: Rhythmic entrainment and the motor system.” Frontiers in Psychology, 5:1185. doi:10.3389/fpsyg.2014.01185
[14] Bood, R.J., Nijssen, M., van der Kamp, J., & Roerdink, M. (2013). “The power of auditory-motor synchronization in sports: Enhancing running performance by coupling cadence with the right beats.” PLOS One, 8(8):e70758. doi:10.1371/journal.pone.0070758
[15] Frontiers in Psychology. (2025). “Beta frequency binaural beats combined with preferred music enhance combat performance and recovery responses in amateur kickboxers: A randomized crossover trial.” Frontiers in Psychology. doi:10.3389/fpsyg.2025.1636856
[16] Frontiers in Psychology. (2014). “Auditory driving of the autonomic nervous system: Listening to theta-frequency binaural beats post-exercise increases parasympathetic activation and sympathetic withdrawal.” Frontiers in Psychology, 5:1248. doi:10.3389/fpsyg.2014.01248
[17] Nature Scientific Reports. (2024). “Binaural beats at 0.25 Hz shorten the latency to slow-wave sleep during daytime naps.” Scientific Reports. doi:10.1038/s41598-024-76059-9
[18] Lin, Y.T., et al. (2024). “Examining the effects of binaural beat music on sleep quality, heart rate variability, and depression in older people with poor sleep quality in a long-term care institution: A randomized controlled trial.” Geriatrics & Gerontology International. doi:10.1111/ggi.14827
[19] Ingendoh, R.M., Posny, E.S., & Heine, A. (2023). “Binaural beats to entrain the brain? A systematic review of the effects of binaural beat stimulation on brain oscillatory activity, and the implications for psychological research and intervention.” PLOS One. doi:10.1371/journal.pone.0286023
Document version: April 2026 Contact: deepmade.ai