Surface electromyography, usually abbreviated as sEMG or SEMG, is a technique that measures the electrical activity your muscles produce when they contract, using sensors placed on the skin rather than needles inserted into the tissue. Every time a muscle fiber fires, it generates a tiny electrical signal that travels outward through the surrounding tissue to the skin surface, where electrodes can pick it up. The resulting recording looks like a noisy, oscillating waveform, but it encodes a surprising amount of information about how hard a muscle is working, when it switches on and off, and whether it is starting to fatigue. sEMG is used across rehabilitation medicine, sports science, ergonomics, prosthetic limb control, and even silent speech recognition, making it one of the more versatile measurement tools in applied physiology.
Where the Signal Comes From
Muscle contraction begins when your brain sends a command down a motor neuron. That neuron branches out to connect with a group of muscle fibers, and together the neuron and its fibers form a motor unit. When the motor neuron fires, electrical impulses called action potentials are generated at the junction between nerve and muscle, then propagate along all the fibers in that motor unit toward the tendons at each end. The sum of all those individual fiber action potentials within a motor unit creates what researchers call a motor unit action potential. sEMG picks up the combined electrical activity of many motor units firing at once, recorded through the skin at varying distances from the source.1PubMed Central. Current developments in surface electromyography
Because the signal has to travel through layers of skin, fat, and connective tissue before reaching the electrode, it arrives somewhat blurred and weakened compared to what you would see with a needle electrode sitting right inside the muscle. That blurring is the fundamental trade-off of sEMG: you get a painless, non-invasive recording, but you sacrifice spatial precision. The electrode “sees” activity from a broad region rather than from a single motor unit, and the deeper or smaller the muscle, the harder it is to isolate its signal from the surface.
What Subcutaneous Fat Does to the Signal
One of the most practical concerns for anyone using sEMG is body composition. The layer of fat between the muscle and the skin acts as a low-pass filter, attenuating the signal before it ever reaches the electrode. Modeling work has shown that even a modest fat layer of about 3 mm reduces the root mean square amplitude of the signal by roughly a third. At 9 mm the signal drops by about 80%, and at 18 mm it is almost entirely lost, with a 90% reduction. To make matters worse, the thicker the fat layer, the more “cross-talk” appears, meaning the electrode starts picking up activity from neighboring muscles rather than from the one directly beneath it.2PubMed Central. The effect of subcutaneous fat on myoelectric signal amplitude and cross-talk This is why sEMG recordings from people with higher body fat percentages need especially careful interpretation, and why researchers often measure skinfold thickness at electrode sites.
Electrode Placement and Standardization
Getting useful data from sEMG depends heavily on putting the electrodes in the right spot. Shift them by a centimeter or two and you might record a completely different muscle, or land over a tendon where there is little electrical activity to pick up. To address this, a European research initiative called SENIAM developed detailed placement guidelines for 27 different muscles, covering electrode type, size, orientation, and exact anatomical landmarks. These recommendations became the most widely cited reference in the field.3Elsevier / PubMed Central. Development of recommendations for SEMG sensors and sensor placement procedures A complementary set of standards from the International Society of Electrophysiology and Kinesiology (ISEK) addresses signal acquisition, processing, and reporting, and the two frameworks together form the backbone of reproducible sEMG research.4CrossRef. Standardized frameworks for myoelectric research: SENIAM and ISEK guidelines, applications, and ethical perspectives
That said, SENIAM guidelines are not perfect for every muscle. A recent study evaluating electrode placement on the triceps surae (the calf muscles) found that SENIAM’s recommended positions for the lateral gastrocnemius landed very close to the edge of the muscle boundary, and in four participants the electrodes ended up outside the muscle entirely. An expert using manual palpation placed them more centrally and avoided that problem. For the soleus, SENIAM placements sat uncomfortably close to the tibial bone. The medial gastrocnemius fared better under both methods.5PubMed Central. Assessment of common electrode placement methods for surface electromyography recordings of the triceps surae muscles The practical lesson is that standardized protocols give you a strong starting point, but individualized adjustments guided by anatomy and palpation improve accuracy, especially in muscles with variable geometry across people.
Cleaning Up the Signal
Raw sEMG is messy. The signal you want, the electrical activity of the target muscle, competes with noise from several sources. Power line interference at 50 or 60 Hz (depending on your country) is the most common contaminant, creating a persistent hum through the recording. Motion artifacts from the electrodes shifting on the skin add low-frequency disturbance. And electrical activity from the heart (the ECG signal) can bleed into recordings from trunk muscles. Removing these contaminants while preserving the true muscle signal is a core part of any sEMG workflow.6Europe PMC. Reducing Noise, Artifacts and Interference in Single-Channel EMG Signals: A Review
For power line noise specifically, a simple notch filter at 50 or 60 Hz is the traditional fix, but it can also remove legitimate muscle signal energy at that frequency. More sophisticated approaches use wavelet-based methods to estimate and subtract the noise harmonics without distorting the underlying EMG. One such technique achieved correlation coefficients of about 0.99 between the cleaned and original pure signals in testing, preserving nearly all the signal energy while suppressing the interference.7PubMed Central. Power line interference filtering on surface electromyography based on the stationary wavelet packet transform
Cross-Talk and How to Reduce It
Cross-talk is the unwanted pickup of electrical signals from muscles adjacent to the one you are trying to measure. It is arguably the biggest validity threat in sEMG, because it can make it look like a muscle is active when it is actually silent, or inflate correlations between muscles that are not truly coordinated. In standard monopolar recordings, about 80% of motor units generate detectable cross-talk in neighboring muscle channels. Spatial filtering techniques reduce this substantially: single differential recording (the standard bipolar electrode pair) drops the proportion to around 50%, and double differential filtering brings it down further to about 42%.8PubMed Central. Surface EMG cross talk quantified at the motor unit population level for muscles of the hand, thigh, and calf
The double-differential approach works by using a three-electrode array and computing a second spatial derivative, which narrows the recording’s “field of view” and improves selectivity by up to six-fold compared to simpler methods.9Elsevier / Clinical Neurophysiology. A convenient method to reduce crosstalk in surface EMG High-pass temporal filtering, by contrast, does not help with cross-talk because the contaminating signals occupy the same frequency range as the genuine ones. If you are recording from small or closely packed muscles, like those in the forearm or the lower leg, spatial filtering is not optional; it is essential for trustworthy results.
Why Raw Numbers Are Meaningless Without Normalization
One thing that trips up newcomers to sEMG is assuming the raw voltage values have inherent meaning. They do not. The amplitude of an sEMG signal depends on electrode size, skin impedance, how much gel you used, the exact placement, subcutaneous fat thickness, and even how recently the skin was shaved. Comparing a raw amplitude of 200 microvolts from one person’s biceps to 300 microvolts from someone else’s tells you essentially nothing about who is working harder. Even comparing the same person’s left and right sides in raw values is unreliable if the electrode conditions differ at all.10PubMed Central. The importance of normalization in the interpretation of surface electromyography: a proof of principle
The solution is normalization: expressing the recorded signal as a percentage of some reference contraction. The most common approach is the maximal voluntary isometric contraction (MVIC), where you ask someone to push as hard as they can against a fixed resistance, record the peak sEMG during that effort, and then report all subsequent activity as a percentage of that peak. An alternative is using a standardized functional task, like single-leg stance for lower limb muscles. Both methods have been shown to be reliable, with intraclass correlation coefficients above 0.80, though the MVIC method tends to produce lower variability between repeated tests.11PubMed Central. Reliability and interpretation of single leg stance and maximum voluntary isometric contraction methods of electromyography normalization Without normalization, sEMG data is difficult to interpret meaningfully and nearly impossible to compare across sessions, people, or studies.
Surface Versus Intramuscular EMG
sEMG is the right tool for many applications, but it is not the right tool for all of them. Intramuscular EMG, which uses fine-wire or needle electrodes inserted directly into the muscle, can record from deep muscles that surface electrodes simply cannot reach. It also provides much sharper spatial resolution, making it possible to isolate the activity of individual motor units.
A comparison of sEMG and intramuscular EMG techniques concluded that sEMG is best suited for short-duration, isometric assessments of large, superficial muscles, while intramuscular methods (particularly multi-motor-unit recording) are better for studying deep or small muscles during both static and dynamic activities. Intramuscular EMG is also less affected by signal cancellation and motion artifacts.12SpringerLink. A comparison of electromyography techniques: surface versus intramuscular recording In practical terms, direct comparisons of the two methods during walking have found that for some calf muscles, like the medial gastrocnemius, the surface and intramuscular recordings match well at all speeds. For others, like the tibialis anterior, the two methods diverge substantially during mid-stance and late stance across walking speeds.13Frontiers in Physiology. Comparing Surface and Fine-Wire Electromyography Activity of Lower Leg Muscles at Different Walking Speeds The disagreements matter because they mean sEMG can misrepresent the timing or intensity of certain muscles’ contributions to movement. Knowing which muscles are “sEMG-friendly” and which are not is part of designing a valid study.
Detecting Muscle Fatigue
One of the most popular uses of sEMG is tracking fatigue. When a muscle tires, the electrical signals it produces change in characteristic ways: the frequency content shifts downward (the power spectrum compresses toward lower frequencies), and the amplitude tends to increase as the nervous system recruits more motor units to compensate for failing ones. Tracking the shift in median frequency of the sEMG power spectrum is the classic method for detecting fatigue, and it has been widely applied in both laboratory and workplace settings.14PubMed Central. The relationship between EMG median frequency and low frequency band amplitude changes at different levels of muscle capacity
That said, median frequency is not as clean a fatigue indicator as textbooks sometimes imply. It does not always behave consistently, and it can struggle to distinguish reliably between fatigued and non-fatigued states, particularly at low levels of fatigue.15PubMed Central. Non-invasive detection of low-level muscle fatigue using surface EMG with wavelet decomposition Researchers have explored wavelet-based decomposition methods to extract more sensitive fatigue markers from the time-frequency domain, with some success. In occupational settings, sEMG-based fatigue monitoring has been used during manual handling in factories, at supermarket checkout stations, and during endoscopic surgery, demonstrating the technique’s versatility for real-world ergonomic assessment.16Elsevier. Electromyographical indication of muscular fatigue in occupational field studies
High-Density sEMG and Motor Unit Decomposition
A major advancement in recent years is high-density surface EMG, or HD-sEMG. Instead of a pair of electrodes over a single muscle, HD-sEMG uses a grid of dozens or even hundreds of closely spaced electrodes, typically a few millimeters apart. This dense sampling creates a spatial map of the electrical field across the skin surface. With the right algorithms, researchers can then decompose that map into the firing patterns of individual motor units, something that used to require needle electrodes.17Wiley Online Library. Unlocking the full potential of high-density surface EMG: novel non-invasive high-yield motor unit decomposition
HD-sEMG decomposition has been applied to study the neural control of movement in both healthy people and clinical populations. In stroke survivors, for example, researchers have used it to simultaneously record motor unit discharge patterns from multiple muscles in the arm’s flexion synergy, including the deltoid, biceps, and wrist flexors, during isometric tasks. This gives clinicians a non-invasive window into how the nervous system reorganizes its control strategies after brain injury.18PubMed Central. High-density surface EMG decomposition allows for recording of motor unit discharge from proximal and distal flexion synergy muscles simultaneously in individuals with stroke
Prosthetic Limb Control
One of the most visible applications of sEMG is in controlling myoelectric prostheses. The idea is straightforward: electrodes on the residual limb detect the sEMG patterns associated with different intended movements, and a classification algorithm translates those patterns into commands for the prosthetic hand or arm. Pattern recognition approaches allow for multi-function control, so a user can potentially switch between gripping, pointing, rotating, and other hand postures.
The technology works well for many people with acquired amputations, but it is less predictable for those with congenital limb differences. A preliminary study of congenital transradial amputees found that overall classification accuracy on the residual limb averaged about 52%, which was roughly 41 percentage points lower than accuracy on the intact limb. Some individuals performed well enough for practical use, but others did not, likely because the residual muscles had never developed the movement-specific activation patterns that the algorithms rely on.19PubMed Central. Pattern recognition control of multifunction myoelectric prostheses by patients with congenital transradial limb defects: A preliminary study Machine learning models continue to improve: recent work using deep learning architectures can estimate continuous hand joint angles from sEMG in real time, moving beyond discrete gesture classification toward fluid, natural-feeling control.20Nature. A parallel and efficient transformer deep learning network for continuous estimation of hand kinematics from electromyographic signals
Pelvic Floor Assessment
A less well-known but growing application of sEMG is evaluating pelvic floor muscles. Unlike the limb muscles people typically associate with EMG, the pelvic floor is assessed using specialized intravaginal or intrarectal sensor probes that sit against the mucosal surface. The recordings help clinicians distinguish between different pelvic floor conditions. In one comparative study, women with pelvic floor dyssynergia (muscles that contract when they should relax) showed higher sEMG amplitudes during sustained contractions compared to both healthy women and those with urinary incontinence, along with impaired relaxation after contraction. Women with urinary incontinence, meanwhile, showed lower contraction amplitudes than the healthy group.21PMC. Surface Electromyography Characteristics of Pelvic Floor Muscles in Healthy Women, Pelvic Floor Dyssynergia, and Urinary Incontinence: A Retrospective Comparative Study
Pelvic floor sEMG has shown good reliability in both healthy women and women with dysfunction, though standardized normalization methods remain important for comparing across patients.22Elsevier. Reliability of pelvic floor muscle electromyography tested on healthy women and women with pelvic floor muscle dysfunction Recent work has applied AI to establish more accurate reference ranges for pelvic floor sEMG, outperforming the traditional Glazer protocol for diagnosing pelvic floor dysfunction. The AI-derived reference ranges achieved diagnostic accuracy (measured by area under the curve) of about 0.81 compared to the Glazer protocol’s 0.76 on the same test data.23The Lancet. AI-based pelvic floor surface electromyography reference ranges and high-precision pelvic floor dysfunction diagnosis
Workplace Ergonomics and Back Loading
Beyond the clinic, sEMG is becoming a tool for workplace safety. The ability to monitor muscle activity during actual job tasks, rather than just in a lab, makes it valuable for identifying movements or postures that overload the body. One emerging approach combines sEMG-driven musculoskeletal models with wearable sensors to estimate compressive and shear loads at the lower back during lifting. In a study evaluating lifting tasks at different risk levels, personalized models driven by sEMG data produced back-loading estimates that tracked well with the risk levels defined by established ergonomic guidelines, though the actual compressive loads exceeded the thresholds those guidelines proposed.24Frontiers in Bioengineering and Biotechnology. Assessing low-back loading during lifting using personalized electromyography-driven trunk models and NIOSH-based risk levels The long-term vision is wearable sEMG garments that continuously monitor back loading in factory or warehouse settings, flagging risky movements before injury occurs.
Textile Electrodes and Wearable sEMG
Traditional sEMG requires adhesive gel electrodes stuck to the skin, which dry out over time, irritate the skin with prolonged use, and make repeated daily monitoring impractical. Textile-based electrodes aim to fix this by embedding conductive materials directly into clothing. A smart leg sleeve incorporating e-textile sensors has been developed to record sEMG from the tibialis anterior and lateral gastrocnemius without gel or adhesive, making it far more practical for long-term ambulatory monitoring.25Europe PMC. Measuring Surface Electromyography with Textile Electrodes in a Smart Leg Sleeve
How well do these fabric electrodes actually work? Testing of screen-printed and inkjet-printed dry textile electrodes found that, at a 20 mm diameter and about 10% stretch for good skin contact, the signal-to-noise ratios came within a few decibels of conventional wet electrodes, around 23 to 25 dB compared to 26 dB. When paired with a small Bluetooth-enabled amplifier, the system also reduced motion artifacts and eliminated the cable tangle of traditional setups.26Cambridge University Press. Feasibility assessment of textile electromyography sensors for a wearable telehealth biofeedback system The technology is not yet a plug-and-play replacement for clinical-grade systems, but it is approaching the performance threshold where continuous home-based sEMG monitoring becomes realistic.
Silent Speech Recognition
Perhaps the most unexpected application of sEMG is reading words that are never spoken aloud. When you silently mouth a word, the muscles of your face, jaw, and neck still activate in patterns specific to that word, even though no sound comes out. Researchers have developed face- and neck-worn sEMG sensor arrays coupled with recognition algorithms that can decode silently mouthed words and phrases entirely from muscle signals.27IOPscience. Development of sEMG sensors and algorithms for silent speech recognition The intended uses range from restoring communication for people who have lost their voice (after laryngectomy, for instance) to covert communication for military personnel and private phone conversations in public. The accuracy is still limited to relatively small vocabularies, but the concept demonstrates just how far sEMG has traveled from its origins as a simple muscle activity recorder.