When we set out to test Sonopeace, we made a deliberate choice: we would hold ourselves to the same standard used in pharmaceutical research. Not because we had to, but because anything less would not give us answers we could trust.
This article walks through the full design of our clinical study: why we chose the methods we did, how we protected against bias, what the data collection looked like, and what we found. If you have ever wondered what it actually takes to run a rigorous sleep study, this is for you.
Why a Double-Blind, Placebo-Controlled, Cross-Over Design
Clinical trials come in many forms, and each design choice involves trade-offs. We chose a combination of three features that, together, represent the gold standard for evaluating efficacy in a small-sample study.
Double-blind means that neither the participants nor the researchers running the study knew who was receiving Sonopeace and who was receiving the placebo at any given time. This is important because expectation shapes experience. If a participant believes they are using the "real" device, they may unconsciously report better sleep. If a researcher knows which group a participant is in, they may subtly influence interactions. Blinding removes both sources of bias.
Placebo-controlled means every participant used a control device that looked, felt, and sounded like the real thing. Without a placebo arm, any improvement could be attributed to the novelty of sleeping with a new device, the comfort of the shruti drone sound, or simply the passage of time. A placebo control isolates the specific effect of the Sonopeace frequency profile from everything else happening during the study.
Cross-over means every participant served as their own control. Rather than splitting people into two separate groups (one that only gets treatment, one that only gets placebo), a cross-over design gives each person both conditions in sequence. This is particularly powerful in sleep research, where individual variation is enormous. Two people can have dramatically different sleep architectures, sensitivities, and baseline severity. By comparing each participant to themselves, rather than to a group average, the cross-over design cuts through that noise and reveals the actual treatment effect with fewer participants.
The study was approved by Sterling Institutional Review Board under protocol number 13207. Statistical analysis was authored by Dr. Peter A. McNair, Dr.Med.Sc., MD, of The Mereon Legacy CIC, with advisory oversight from Dr. Michele Okun, Director of Clinical and Biobehavioral Research at the University of Colorado at Colorado Springs.
How SleepScore Labs Ran the Study
An independent commercial sleep laboratory managed all participant communication, device distribution, and data collection. No one at the lab knew which participants were assigned to which protocol. This organizational separation is a critical safeguard: it ensures the people interacting with participants cannot inadvertently influence outcomes.
Each participant received a package containing two sealed audio devices, one marked for the first listening period and one for the second. The devices were identical in appearance and volume. The only difference was the encrypted sound file inside. Participants received written instructions before each period and completed questionnaires daily throughout the seven-week study.
The entire study ran from July 7 to August 25, 2025, with all participants starting and ending on the same dates. This synchronized timeline eliminated seasonal and environmental variation between groups.
Building a Believable Placebo
Designing a convincing placebo for a sound-based intervention is one of the harder problems in clinical trial design. Unlike a sugar pill, a sound device has to actually produce something. The participant needs to hear and feel it working. Otherwise, blinding fails.
The Sonopeace device delivers eight sequential frequencies in the gamma range via bone conduction, played through a transducer positioned beneath the pillow or mattress. The frequencies cycle over a 37-minute period and repeat throughout an eight-hour sleep session. Between each frequency, a continuous shruti drone (an Indian drone instrument) bridges the transitions.
For the placebo device, we used the shruti drone alone, without the Sonopeace frequency profile layered on top. The shruti volume was set relatively high on both devices to mask any perceptible difference between them. Both devices were calibrated to the same total sound volume before shipping. From the participant's perspective, both devices produced a continuous, pleasant drone sound delivered through bone conduction. They had no reliable way to distinguish one from the other.
This approach is standard practice in well-designed sound intervention studies. The key principle is that the placebo must be credible enough that participants genuinely do not know which condition they are in. Post-study data confirmed this was the case.
Participant Selection: From 469 to 35
Rigorous selection criteria are what separate a meaningful clinical result from a noisy one. We started with 469 individuals who completed an eligibility questionnaire from a database of people interested in participating in sleep research.
Of those 469, 414 were immediately excluded based on inclusion and exclusion criteria. Another 10 were excluded through first-come-first-serve selection, leaving 45 enrolled participants.
To qualify, participants had to be adults aged 40 to 60, living in the United States, in good general health, with self-reported difficulty falling asleep or staying asleep. They also needed a score below 5 on the STOP-BANG screener, which rules out undiagnosed obstructive sleep apnea.
The exclusion list was extensive and intentional. We excluded anyone with a diagnosed clinical sleep disorder, untreated medical conditions affecting sleep (such as hyperthyroidism, asthma, or diabetes), untreated psychological conditions affecting sleep (such as clinical depression or bipolar disorder), prescription or over-the-counter sleep medications (including melatonin), medications for other conditions that affect sleep (such as anxiety or ADHD medications), heavy alcohol consumption, recreational drug use, hearing loss, shift work, and significant travel during the study period. Pregnant or nursing women and parents of infants under one year were also excluded.
These criteria served two purposes. First, they ensured that observed changes could be attributed to the intervention rather than to medication interactions, untreated disorders, or lifestyle confounds. Second, they created a study population that mirrors the most common Sonopeace customer: someone with genuine, persistent sleep difficulty who is not managing a diagnosed clinical condition.
Of the 45 enrolled, 10 did not complete the study for various reasons, including device issues, family or medical circumstances, hospitalization from a traffic accident, and loss of contact. The final analysis included 35 completers: 18 in Protocol A (placebo first, then Sonopeace) and 17 in Protocol B (Sonopeace first, then placebo).
At baseline, participants had a mean PSQI score of 7.7 (the clinical threshold for poor sleep is 5), a mean ISI score of 13.5 (borderline moderate clinical insomnia), and a mean PROMIS T-score of 58 (population norm is 50). These were not people with minor sleep complaints. They were genuinely struggling.
How We Collected Data: Objective and Subjective
Sleep research that relies on only one type of measurement is incomplete. People are notoriously unreliable reporters of their own sleep, often underestimating how much they slept or overestimating how long they were awake. But objective sensors alone miss something important: how someone feels about their sleep matters for their quality of life, regardless of what the numbers say.
We used both.
Objective measurement: SleepScore Max
Every participant was given a SleepScore Max, a non-contact radiofrequency sensor placed on the nightstand beside the bed. It uses ultra-low-power radio waves to detect respiratory patterns and body movement, similar to low-energy radar. From these signals, proprietary algorithms derive sleep stages (wake, light, deep, and REM), sleep onset latency, number of awakenings, and total sleep time.
The SleepScore Max is the same technology used in clinical sleep research settings. It requires no wearable device, no headband, and no participant manipulation of data. It simply records what happens while the participant sleeps. This passive, contactless approach avoids the "first-night effect" that wrist-worn or head-worn trackers can introduce, where the device itself disrupts the sleep it is trying to measure.
Subjective measurement: four validated questionnaires
At the end of each study period (baseline, period 1, and period 2), participants completed four standardized clinical instruments:
- PROMIS Sleep-Related Impairment Short Form 8a (primary outcome): measures how sleep problems affect daily functioning. Developed by the National Institutes of Health, PROMIS is one of the most widely validated patient-reported outcome measures in sleep research.
- Pittsburgh Sleep Quality Index (PSQI): a 19-item questionnaire that assesses overall sleep quality across seven components, including duration, disturbance, and daytime dysfunction. A score above 5 indicates clinically poor sleep.
- Insomnia Severity Index (ISI): a seven-item measure of insomnia severity, with established thresholds for clinically meaningful response (6 or more points of improvement) and remission (score below 8).
- Epworth Sleepiness Scale (ESS): measures daytime sleepiness. This was included not to show improvement but to confirm safety. If Sonopeace were causing sedation rather than genuine sleep quality improvement, ESS scores would increase.
In addition to these period-level questionnaires, participants completed brief daily questionnaires each morning, reporting on perceived sleep quality, how rested they felt, and their experience with the device. This produced a rich, granular picture of sleep experience over all 49 nights of the study.
The 7-Week Timeline
The study spanned seven weeks, divided into three distinct periods:
- Period 0 (weeks 1 through 3): Three-week baseline observation. No device. Participants simply slept as normal while the SleepScore Max recorded their sleep and they completed daily questionnaires. This established each person's individual sleep baseline.
- Period 1 (weeks 4 and 5): Two-week first listening period. Protocol A participants used the placebo device. Protocol B participants used the Sonopeace device.
- Period 2 (weeks 6 and 7): Two-week second listening period. Protocol A participants switched to the Sonopeace device. Protocol B participants switched to the placebo device.
The three-week baseline is longer than is typical in many sleep studies. We chose this duration to get a stable, reliable picture of each participant's natural sleep patterns before introducing any intervention. Sleep varies night to night. A single week of baseline data can be misleading. Three weeks provides enough data to establish a trustworthy individual reference point.
The two-week treatment periods were set based on sample size calculations targeting a minimal relevant difference of 10 to 15 percent, with alpha at 0.05 and power at 0.8. This duration is long enough to detect meaningful changes while short enough to maintain participant compliance across the full study.
Key Results
The full statistical analysis runs to hundreds of pages. Here are the findings that matter most.
Primary outcome: 106% more improvement in sleep-related impairment
PROMIS, our primary outcome measure, showed the study's strongest result. Participants who used Sonopeace first (Protocol B) showed 106% more improvement in sleep-related impairment compared to those who used the placebo first (Protocol A). This difference was statistically significant (p = 0.020, two-tailed) with a medium-to-large effect size (Cohen's d = 0.72).
In the direct Period 1 comparison, before any carryover effects could complicate the picture, Sonopeace produced significantly greater improvement than placebo (p = 0.032, one-tailed; Cohen's d = 0.60).
Objective sleep improvements
The SleepScore Max data showed statistically significant improvements in deep sleep duration (p = 0.026), with a between-group difference of nearly 8 additional minutes of deep sleep per night in Period 1 (p = 0.017). Overall sleep quality scores improved significantly (p = 0.006), and total sleep time increased by up to 20 minutes per night (p = 0.005). REM sleep duration also increased significantly (p = 0.003).
Insomnia severity
43% of participants achieved a clinically meaningful reduction in insomnia severity (6 or more points on the ISI). 37% achieved full remission from clinical insomnia, meaning their ISI score dropped below the clinical threshold entirely. The overall ISI improvement was 4.17 points (p < 0.001) with a very large effect size (Cohen's d = 1.12).
Sleep quality
37% of participants achieved clinically meaningful improvement in sleep quality as measured by the PSQI (3 or more points of improvement). The group moved from a mean score of 7.74 (poor sleep) toward 5.83 (approaching normal range), with the change reaching high statistical significance (p < 0.001).
No sedation, no dependency
The Epworth Sleepiness Scale showed no increase in daytime sleepiness (p = 0.878). This is an important safety finding. It confirms that Sonopeace improves sleep quality through genuine sleep architecture improvement, not through sedation. There is no "hangover effect" the next morning.
Perhaps most intriguingly, participants who used Sonopeace first and then switched to the placebo device maintained their improvements. PROMIS scores continued to improve even after Sonopeace was discontinued. Sleep quality scores in the Sonopeace-first group remained significantly higher than the placebo-first group at study end (p = 0.021). The report concludes that "Sonopeace benefits appear durable... users may not need continuous nightly use."
What the Placebo Effect Tells Us
We want to be transparent about something: participants in the placebo group also improved. Several subjective and objective measures showed gains during the placebo period. This is common in sleep research. The act of participating in a study, paying attention to sleep habits, and using any device before bed can produce real benefits.
This is exactly why a placebo control exists. The question was never "does sleeping with a device help?" It was "does the Sonopeace frequency profile produce benefits beyond what the device experience alone provides?" The answer, particularly on the PROMIS primary outcome and the objective deep sleep and sleep quality measures, is yes.
What Comes Next
This study was designed as a first step: a proof-of-concept trial to determine whether Sonopeace produces measurable, statistically significant effects on sleep quality in a real-world home setting. It achieved that goal.
We also recognize the limitations. Thirty-five completers is a meaningful sample for a cross-over design, but larger studies will strengthen confidence in the findings. The participant pool was predominantly female (89%) and within a specific age range (40 to 60). Future research should include broader demographics. And while the carryover effect is a promising signal, understanding optimal usage patterns, whether nightly use, intermittent use, or periodic "booster" sessions, will require longer-term follow-up.
We are committed to continued research. The data from this trial is being prepared for peer review, and planning for follow-up studies is underway. We believe that transparent, rigorous evidence is the foundation of trust, and we intend to keep building on it.
For a complete overview of our research program, visit Our Studies.