TEST SPEAKERS & MICROPHONE ONLINE

Connect the microphone before opening the test. For a USB microphone or audio interface, use a direct computer port during diagnosis instead of an unpowered hub. For a 3.5 mm headset, confirm that the plug matches the computer jack; separate headphone and microphone plugs may need a four-pole headset adapter. Charge wireless headsets and verify that they are connected as an audio device rather than merely paired.

Set hardware gain conservatively. Place a headset boom slightly to the side of your mouth so breath does not hit the capsule. Position a desk microphone roughly 10 to 20 centimeters away, with the correct side facing you. Turn off any physical mute switch, but keep speaker volume low until the speaker test begins.

• Close unused recording, meeting, streaming, and voice-chat applications.
• Use the same cable, USB port, headset profile, and room planned for the real session.
• Reduce obvious noise sources such as fans, open windows, keyboard vibration, and desk contact.

Press Start microphone and allow access when the browser asks. Modern browsers require a deliberate permission decision because a microphone is a sensitive input. Device labels may remain generic until access is granted. After the stream opens, choose the intended microphone from the input list. A laptop can expose its internal array, a webcam microphone, a USB interface, a display microphone, and a Bluetooth headset at the same time.

Speak while switching only one input at a time. The correct choice should react immediately and should change when you tap near that physical device. Do not assume that System default matches a meeting application: Zoom, Teams, Meet, Discord, an operating-system mixer, and the browser can each remember a different input.

Speak in a normal voice for at least ten seconds. Include quiet speech, your expected average level, and one intentionally louder sentence. Watch the waveform and input meter, but judge the session peak and clipping result rather than chasing a constantly high meter. Digital audio needs headroom; a healthy spoken peak often sits below 0 dBFS instead of touching the maximum.

Record a 15 to 30 second sample using words with different sounds, short pauses, and one keyboard or mouse action. Play it back through headphones when possible. Listen for clear consonants, steady volume, hiss, hum, room echo, pumping, clipped syllables, wireless dropouts, and aggressive noise reduction. This playback is the closest browser-local approximation of what another participant or recording application may receive.

After capture works, lower the system volume and use the speaker controls. Play Both first, then Left and Right. The short tone should be clean, steady, and centered when both channels play. Left should be silent on the right side, and Right should be silent on the left side. This step verifies output routing; it does not use the microphone recording.

If you can record but cannot hear playback, the microphone is not the likely cause. Select another output where supported, inspect the operating-system default speaker, unmute the browser tab, and check per-application volume. Separating these paths prevents unnecessary microphone driver changes when the actual problem is a speaker, headphone, HDMI, Bluetooth, or mixer route.

The meter uses dBFS, or decibels relative to digital full scale. Zero dBFS is the maximum representable digital level; values are normally negative. Current level follows recent energy and falls during silence. Session peak retains the loudest observed moment. A reading near -60 dBFS is effectively quiet for this display, while speech commonly moves through a broad range depending on gain, distance, browser processing, and microphone sensitivity.

There is no universal perfect number for every call platform. As a practical starting point, normal speech that often reaches roughly -24 to -10 dBFS with occasional peaks below about -6 dBFS usually leaves useful headroom. Very low readings may sound weak after compression. Repeated peaks near 0 dBFS can flatten transients and create harsh distortion. Confirm with playback because automatic gain can make a low visual reading sound acceptable or make background noise unnaturally loud.

The noise-floor estimate follows quieter moments when the level drops below typical speech. It can include computer fans, heating, traffic, electrical hum, microphone self-noise, interface preamp noise, and room reflections. A more negative value is quieter. The estimate becomes more useful when you remain silent for several seconds; continuous speech gives the test little quiet material to analyze.

Clipped samples are waveform points extremely close to digital maximum. A tiny isolated count can come from a cable connection or abrupt handling noise, but repeated clipping during speech indicates too much gain or too little distance. Reduce hardware input gain first. Then lower the operating-system input level, move slightly farther from the microphone, or disable automatic gain if it is pushing loud speech upward. Software cannot reconstruct peaks after they have been clipped.

Sample rate is how many audio samples the active track reports per second. Voice systems commonly use 48 kHz, while 44.1 kHz is common in music workflows. A higher number does not automatically mean clearer speech; the microphone, analog electronics, room, encoding, and destination service often matter more. Channel count is frequently one for microphones even when a headset plays stereo. Some interfaces expose two channels because they combine two hardware inputs.

Sample size describes reported bit depth when the browser provides it. Latency is a target or browser-reported processing delay, not an end-to-end measurement of a call. This page does not claim to measure network delay, service buffering, acoustic echo round-trip time, or the delay heard by another person. Missing values mean the browser or driver did not expose them, not that the feature is absent or broken.

Echo cancellation tries to remove sound from speakers that re-enters the microphone. Noise suppression tries to reduce steady background sound. Automatic gain control raises or lowers input level to keep speech in a usable range. These features are valuable for calls, but they can alter music, ambience, soft consonants, keyboard sounds, and natural dynamics. The browser and operating system may implement them differently from the final application.

Use processed settings for a realistic meeting check and disable them when evaluating the raw microphone or recording music. After changing an option, restart the microphone so a new stream requests that setting. The Active stream details show what the track reports, but a browser may ignore unsupported constraints or apply processing in a layer that is not fully described by the exposed value.

Before you test speakers, lower the operating-system volume and remove headphones from your ears if you are unsure of the current level. The page limits tone amplitude and offers a modest range, but the final loudness still depends on the system mixer, amplifier, interface, and speaker sensitivity. Pure sine tones concentrate energy at one frequency and can feel more intense than speech or music.

Start with 440 Hz through Both. Increase volume only enough to hear a stable tone. Then play Left and Right. Use 1000 Hz to check a clear midrange reference and 2000 Hz only at a comfortable low level. This is a routing and basic reproduction check, not a calibrated frequency-response measurement. Avoid long repeated tones at high volume.

With headphones, Both should appear near the center. With a stereo speaker pair, it should appear between the cabinets when you sit centrally. Left should come only from the physical left side and Right only from the physical right side. Reversed channels can result from swapped cables, speakers placed on the wrong sides, an audio-interface routing matrix, or a software channel map.

If one side is silent, exchange cables or earbuds between sides when the hardware allows it. A failure that follows the cable or driver points toward hardware. If both sides work individually but centered audio favors one side, inspect the operating-system balance control and the physical speaker gain controls. Also test speakers with another known source before concluding that a browser is responsible.

Some Chromium-based browsers allow a page to route an audio element to a selected output device. Other browsers play through the operating-system default and may show no useful output names. The menu therefore acts as a convenience where supported, not a promise that every browser can override system routing. If selection does not change the sound, set the desired speaker or headset as the operating-system default and run the speaker test again.

HDMI monitors, docks, USB audio interfaces, Bluetooth devices, and virtual audio cables can silently become the default output. A successful recording with silent playback often means audio is going to one of these destinations. Conversely, a successful speaker test with a flat microphone waveform means output works while input selection, permission, mute, or hardware still needs attention.

The recorded clip links capture and playback in a controlled way. If the live waveform moves and the downloaded clip contains audio on another device, capture is working even if local playback is silent. If test tones are audible but the recording is empty, speakers work and microphone capture does not. If both are audible but a meeting application fails, select the same devices inside that application and inspect its own mute, processing, and permissions.

Do not enable live microphone monitoring through speakers during diagnosis. That creates acoustic feedback and makes echo cancellation part of the test. A short recording followed by playback is safer, easier to interpret, and closer to a normal send-then-receive check. Use headphones when comparing microphones or listening for quiet background noise.

Microphones become more direct and less reverberant as they move closer to the speaker, but very small distances exaggerate breath, mouth noise, bass, and level variation. For a desk microphone, begin around 10 to 20 centimeters away. Speak across the capsule at a slight angle instead of directly into it. Use a pop filter for exposed studio microphones and keep headset booms one or two finger widths from the corner of the mouth.

Confirm the pickup side. Many side-address condenser microphones look symmetrical but reject sound from the rear. Some models offer cardioid, omnidirectional, or bidirectional patterns; cardioid usually suits a single computer user. Avoid tapping the body to identify it at high gain. Speak while rotating the microphone and watch for the clearest, strongest orientation.

Gain amplifies the microphone before or during conversion. Too little gain produces a weak signal that later processing must raise along with noise. Too much gain clips loud words and amplifies room sound. Set hardware gain while delivering the loudest realistic sentence, then leave several decibels of headroom. Do not normalize every quiet breath into a loud event.

Automatic gain is convenient for users who change distance, but it can raise fan noise during pauses and pull loud speech down abruptly. Compare one sample with it enabled and one disabled. Match playback loudness before judging quality; the louder sample often seems better even when it contains more noise or distortion. For music, instruments, and controlled voice recording, stable manual gain is usually easier to evaluate.

Move the microphone away from laptop exhaust, desktop fans, air vents, open windows, and hard reflective walls. Put a desk stand on a soft isolation pad or use a shock mount so keyboard and mouse impacts do not travel through the surface. A directional microphone should point its least-sensitive area toward the dominant noise when its pattern is known.

Soft furnishings reduce high-frequency reflections but do not soundproof a room. Heavy curtains, rugs, books, and upholstered furniture can improve speech clarity by shortening obvious flutter and echo. Low-frequency traffic and building vibration require distance, isolation, or a quieter time of day. Noise suppression can hide steady sound, yet good source acoustics preserve a more natural voice.

Bluetooth headsets often switch from high-quality stereo playback to a hands-free profile when the microphone opens. The hands-free path shares limited bandwidth between input and output, so speaker audio may become narrow, mono, or telephone-like during a microphone test. This behavior is a profile limitation, not necessarily defective earbuds.

For the best playback quality, use the headset only as an output while selecting a separate microphone, or use a wired or USB connection. For calls, the hands-free profile may be acceptable because it supports two-way audio and hardware controls. Test speakers once with the microphone stopped and once with it active to recognize the profile change before an important session.

Prioritize intelligibility, stable level, low echo, and reliable device selection. Enable echo cancellation if sound will play through speakers; use headphones when privacy or feedback matters. Record a sentence at your normal seated posture, then lean back and turn toward a second screen to reveal how quickly level and clarity change. Check that keyboard notes and notifications do not overwhelm speech.

After this browser test passes, open the destination service and choose the same named input and output. Run its preview or test call because service-side codecs, automatic level control, virtual backgrounds, network conditions, and organization policies are outside this page. A local pass narrows the problem; it does not certify the entire call.

Disable browser noise suppression, echo cancellation, and automatic gain when you want to hear the raw capture path. Use headphones, maintain consistent distance, and record room tone before speaking. Listen for mouth noise, plosives, reflections, electrical hum, preamp hiss, chair movement, and computer fans. A clean raw recording gives editing tools more usable material than an aggressively processed one.

This page records a convenient browser format and is suitable for comparison, not as a replacement for a multitrack recorder. It does not expose every interface channel, driver mode, bit depth, clock source, or low-latency buffer setting. Confirm the final sample rate, channel routing, and file format in the production application before a long session.

Test at the same speaking intensity used during action. Include controller, keyboard, mouse, and desk sounds. Noise gates can hide a fan but may cut the first syllable after silence; automatic gain can overreact to shouting. Use the recording to check whether quiet team communication remains understandable and loud reactions stay below clipping.

Streaming software can add filters, virtual devices, and monitoring routes. Begin with the physical microphone directly, then add the virtual or processed input and compare. If the direct device works but the virtual device does not, focus on the routing application rather than reinstalling microphone hardware. Test speakers or headphones separately so game output and chat output are not sent to different devices by mistake.

Compare microphones in the same position with the same phrase, room, gain target, processing state, and playback loudness. Device A sounding louder does not prove it is clearer. Listen for consonant detail, tonal balance, noise, handling sensitivity, room pickup, and consistency when you move. Save clips with descriptive filenames before switching devices.

To isolate a defect, test the microphone on a second port and computer, and test a known-good microphone on the original system. Swap one cable or adapter at a time. If the failure follows the microphone, cable, or interface, hardware becomes more likely. If every microphone fails on one computer, focus on privacy settings, drivers, system routing, policy, or another application holding the device.

Microphone capture requires a secure HTTPS page or localhost and an explicit user decision. Look beside the address bar for a lock, microphone, or settings icon. Change microphone access for this site to Allow, reload, and press Start again. If the site is already allowed, clear its permission and request it again. Private browsing, strict extensions, embedded frames, parental controls, and managed browser policies can block normal prompts.

The operating system has a separate privacy layer. On Windows, verify that microphone access is enabled for the device, applications, and desktop applications. On macOS, check Privacy & Security, then Microphone, and enable the browser. A school or company computer may enforce these settings. Ask an administrator rather than disabling security software globally.

Choose each listed input and speak close to it. Check the headset cable mute, keyboard microphone key, laptop privacy switch, interface mute, and hardware gain knob. Confirm the operating-system input meter moves. If the system meter is also flat, reconnect the device, choose another port, charge the headset, or select the correct audio-interface channel.

Close meeting apps, digital audio workstations, game chat, browser tabs, and vendor utilities that may hold the microphone. Most modern systems share audio devices, but exclusive modes, crashed drivers, or virtual routing tools can still block capture. Restart the browser after closing them. If one microphone fails everywhere while another works, test its cable and another computer.

For low level, move closer, face the correct side, raise hardware gain gradually, and inspect the operating-system input level. Avoid maximizing every control because later gain also raises noise. For distortion, lower hardware gain first, speak off-axis, and increase distance. Watch whether session peak remains near 0 dBFS or clipped samples continue to rise.

For pumping or changing volume, disable automatic gain and compare. For words that disappear, compare noise suppression off and check any gate in a vendor or streaming utility. Crackles can come from a damaged cable, unstable USB hub, overloaded system, interface clock mismatch, or wireless interference. Test a direct port and a simpler signal path before changing software.

A steady high hiss usually reflects gain, microphone self-noise, or a noisy preamp. Low hum can come from power grounding, an analog cable, a charger, or nearby electrical equipment. Disconnect unnecessary analog connections and test a battery-powered laptop briefly. Never defeat protective grounding on mains equipment. Buzz that changes when a cable moves suggests a connector or shielding problem.

Echo in a call usually means speaker sound is reaching a microphone or one participant has duplicate audio paths. Use headphones, lower speakers, and enable echo cancellation. Delayed self-monitoring is normal when audio travels through browser and system buffers. This page uses record-then-playback instead of live monitoring to avoid feedback; use an interface's direct-monitor control when zero-latency monitoring is required.

A successful waveform and recording prove that browser input works. Silent playback points to output. Check the tab mute state, system master volume, per-browser mixer, physical speaker power, headphone connection, and selected output. HDMI displays and Bluetooth devices frequently become default outputs without being audible from your seat.

Use Both, Left, and Right to test speakers. If all are silent, set the desired output as the operating-system default and reload. If test tones work but the recording does not play, try the downloaded file in a system player; browser codec playback support can vary. If another player works, capture is safe and the remaining issue is local browser playback.

Open the application's audio settings and select the exact microphone and speaker names that worked here. Check its own mute button, push-to-talk mode, input sensitivity, noise gate, monitoring device, and permission. Browser-based meeting services can have a separate site permission from this page. Desktop applications have their own operating-system privacy entry.

Disable virtual cables, voice changers, broadcast mixers, and enhancement suites temporarily. Add them back one at a time after the direct hardware path works. For calls, run the service's own test because network upload, codecs, server processing, and account policy are not measured locally. Report that the browser microphone test passed and include the exported settings report when contacting support.

The live waveform, input meter, noise estimate, clipping check, recording, and test tones are produced in your browser. The recording is represented by a temporary local object URL so you can play or download it. This implementation does not upload the audio sample to a server. Closing or reloading the page discards the temporary clip unless you explicitly download it.

The exported JSON report contains browser-visible settings and numeric observations, not recorded sound. The browser still controls the permission indicator and may remember your choice for this origin. Stop the microphone when finished or revoke permission in site settings. The operating system can show which application is currently using the microphone.

Modern microphone capture uses MediaDevices.getUserMedia and requires HTTPS or localhost. MediaRecorder creates the short audio clip, while Web Audio provides the waveform and level analysis. These APIs are broadly available in current browsers, but supported recording containers, device labels, processing constraints, and output selection differ.

Output-device selection is especially inconsistent. Where the browser supports setSinkId, the speaker tone can target a chosen output. Otherwise it uses the system default. Mobile browsers may expose fewer device names and may change audio routing when headphones connect. Keep the browser and operating system current, but do not assume that a missing field means failed hardware.

This is not a calibrated sound-level meter, acoustic analyzer, hearing test, or certification system. dBFS describes digital signal level, not sound pressure level in dB SPL. The noise floor is an estimate from quiet waveform periods, not a laboratory equivalent-input-noise measurement. Speaker tones verify routing and obvious reproduction, not flat frequency response, phase, distortion percentage, or safe listening level.

The page cannot measure another participant's received quality, network packet loss, a conferencing service's final codec, room impulse response, true end-to-end latency, analog preamp specifications, microphone polar pattern, capsule health, or speaker power. Use the results to isolate a browser-visible audio path, then use calibrated equipment or the destination application when the task requires formal measurement.