Your computer is not playing sound because the digital audio signal is being blocked, misrouted, or never generated somewhere along a five-stage chain: the application's audio session, the operating system's audio engine and sample-rate mixer, the device driver and its endpoint assignments, the physical connection or wireless link, and the transducer itself. Most silent-computer cases resolve to one of four causes: the wrong output device is selected, an app is muted in the per-app mixer, a Bluetooth or HDMI endpoint has stolen default status, or a driver or endpoint has crashed into a disabled state. Identifying which stage failed takes about five minutes with the right sequence of checks.
The signal chain: where audio actually dies
Sound on a computer is not one thing but a pipeline, and every stage can fail independently. Understanding the pipeline is what separates a two-minute fix from an hour of reinstalling drivers that were never broken.
The chain begins when an application decodes or generates audio and hands it to the operating system's audio engine. On Windows this is the Audio Device Graph Isolation process hosting the mixer; on macOS it is Core Audio; on Linux it is PipeWire or PulseAudio. The engine resamples everything to a shared mix format, applies per-app and master volume, then passes the combined stream to a driver. The driver translates that stream into a format the hardware endpoint understands, whether that endpoint is a motherboard codec feeding a 3.5 mm jack, a USB audio class device, a Bluetooth radio using the A2DP profile, or a GPU sending audio embedded in an HDMI or DisplayPort signal. Finally the transducer, whether headphones, powered speakers, or TV speakers, converts electrical signal to pressure waves.
The critical insight is that the mixer runs in software and the endpoint runs in hardware, and either side can present as silence. A muted app produces zero samples, so downstream stages have nothing to amplify. A crashed endpoint driver produces a device that looks present but delivers no stream. A Bluetooth link that has dropped to the hands-free profile still connects but sounds absent or garbled. Same symptom, five different layers.
One more mechanism matters: exclusive-mode access. Windows allows applications to take exclusive control of an endpoint, bypassing the mixer entirely. If a conferencing app grabs exclusive mode and then hangs, other applications can appear to play (the progress bar moves) while the endpoint emits nothing, because the mixer's output is being discarded.
Quantified context: how often each layer is the culprit
Silent-audio complaints are among the most common support categories on both major platforms, and the distribution of causes is lopsided. In practice, device-selection and routing errors account for the majority of cases: the system is playing perfectly to an endpoint the user is not listening to. HDMI monitors with built-in speakers, USB headsets, and Bluetooth earbuds all register as valid outputs, and Windows and macOS both switch default devices automatically when a new endpoint appears or a wireless device reconnects. A monitor connected over HDMI can capture default status the moment it wakes from sleep.
Per-app mute states are the second most common layer. Windows 10 and 11 maintain independent volume sliders for every application in Settings > System > Sound > Volume mixer, and these persist across sessions. An app muted once in March stays muted in June, and its internal playback continues silently. macOS has no equivalent per-app mixer by default, which is why Mac routing failures cluster around output-device selection and sample-rate mismatches instead.
Driver and endpoint failures are less frequent but more expensive to diagnose. Windows endpoint corruption typically follows a specific pattern: a driver update, a Windows feature update, or an abrupt shutdown leaves the endpoint enumerated but nonfunctional, and the device shows in Device Manager with a warning icon or shows normally while producing no audio. The sample rate is the quiet variable here. If an application requests 44.1 kHz and the endpoint is configured for 48 kHz, the engine normally resamples transparently, but a misbehaving driver can fail the format negotiation and output silence rather than an error.
The counterintuitive detail professionals rely on: a device that appears in the sound settings list is not evidence that its driver is functioning. Windows enumerates endpoints from registry entries that survive driver crashes. The device is a ghost until the driver stack is rebuilt, and reinstalling the application or rebooting the media player fixes nothing.
Timeline: the typical failure and recovery sequence
| Time from symptom | What is happening | What the user typically observes |
|---|---|---|
| 0 seconds | Application opens a stream to the mixer | Playback starts, progress bar moves |
| 1-3 seconds | Mixer routes to the current default endpoint | No sound; volume icon looks normal |
| 5-30 seconds | User raises master volume | Still silent; nothing changes |
| 1-5 minutes | User checks cables, reconnects headphones | Headphones click, still no sound (wrong endpoint) |
| 5-20 minutes | User restarts the app, then the computer | Sometimes fixed (endpoint reinitialized), often not |
| 20-60 minutes | User reinstalls audio drivers or runs a troubleshooter | Fixed only if the failure was in the driver layer |
| Hours to days | Endpoint corruption persists through updates | Sound returns only after full driver-stack reset |
The pattern to internalize: fixes that take seconds (device selection, per-app mute) address the majority of failures, while fixes that take an hour (driver reinstallation) address the minority. Work the chain from the application outward, not the reverse.
Edge cases that break the standard checklist
The standard advice, check the volume, check the cable, misses several real failure modes that produce identical silence.
Bluetooth profile collapse is the most common edge case. A headset connected for both audio output and microphone can negotiate the hands-free profile instead of the high-quality A2DP profile. The link stays connected, the device shows in settings, and audio either vanishes or degrades to telephone quality. Disconnecting and reconnecting, or disabling the headset's microphone endpoint, forces A2DP renegotiation. Related: Bluetooth devices that reconnect to a different host, such as a phone, leave the computer's endpoint active but silent, because the radio is bonded elsewhere.
HDMI and DisplayPort audio routing is the second trap. A monitor with speakers becomes a default audio endpoint over HDMI, but if the monitor is asleep, switched to another input, or has its own volume at zero, the computer sends a valid stream into a sink that discards it. The fix is not on the computer at all. Similarly, a TV connected over HDMI may have per-input audio settings that mute the PC input.
On macOS, the sample-rate mismatch deserves attention. Audio MIDI Setup lets users set an output device to a specific rate, and a mismatch between the device's selected rate and the app's requested rate normally resamples cleanly. But aggregate devices, virtual audio drivers from conferencing software, and screen-recording utilities can corrupt the aggregate and produce silence until the device is re-selected.
Virtual audio drivers are the third edge case, and the one professionals see most in support tickets. Software installed for streaming, meeting recording, or gaming creates virtual endpoints that hijack the default. The user hears nothing because the system is playing to a virtual cable that routes to a recorder, not to speakers.
Red-flag warnings worth treating as bullets:
Sound works in one app but not another: suspect the application's own output-device setting, not the system. Media players and conferencing apps frequently have internal device selectors that override the system default. Sound stops after a Windows update but the device list looks normal: suspect endpoint corruption; a driver-stack rebuild is the reliable fix. Crackling, popping, or dropouts rather than pure silence: suspect buffer underruns or a mismatched sample rate, not routing. No sound at all and the device is missing from settings: suspect a hardware or firmware fault in the codec or USB controller before reinstalling anything. Sound returns briefly after every reboot and then dies: suspect a background application grabbing exclusive mode or a virtual driver reasserting default status.
Misconceptions that waste diagnostic time
The most expensive misconception is that reinstalling audio drivers fixes most silent computers. It does not, because most failures occur above the driver layer. Driver reinstallation is the correct fix for a narrow class of endpoint corruption, and it is worth doing only after routing and per-app states are excluded.
The second misconception is that the system volume slider reflects reality. On Windows, the master slider and the per-app sliders are independent, and a muted app shows no visual warning anywhere in the volume flyout. On macOS, per-app audio is managed inside each app, so the system slider can be at 100% while a browser tab is muted at the tab level.
The third is that a working microphone proves the output path is fine. Input and output are separate endpoints with separate drivers and separate default assignments. A functioning microphone says nothing about the speaker endpoint.
A professional caveat that contradicts common advice: running the Windows audio troubleshooter is not useless, but its success is concentrated in the device-selection and default-endpoint layers. If it reports the device is working properly, believe it about the driver and stop reinstalling things. The failure is almost certainly upstream in the application or mixer.
Plain-language version: check what the computer thinks it is playing to, then check whether anything is muted along the way, then check the wire or radio, and only then touch the driver.
Practical actions: a diagnostic sequence with thresholds
Work top-down, and stop at the first stage that fails. This sequence resolves the large majority of silent-computer cases without reinstalling anything.
Step 1: verify the endpoint (30 seconds). Open the system sound settings, not the volume flyout, and read the selected output device. Play a known-good test sound, such as a system alert tone, and watch whether the level meter beside the device moves. A moving meter with no audible sound means the signal reaches the endpoint and the problem is physical or in the transducer. A still meter means the problem is upstream.
Step 2: check per-app mixers (1 minute). On Windows, open Volume mixer and confirm no application is muted and each slider is above zero. On macOS, check inside the specific application, then check Audio MIDI Setup for the device's selected sample rate. A tab-level mute in a browser silences that tab only and is invisible to the system.
Step 3: force the correct default (1 minute). Disconnect all non-essential audio endpoints, including Bluetooth devices and HDMI monitors, then reselect the intended output. If sound returns, the failure was automatic device switching, and the long-term fix is to disable automatic default-device switching for endpoints you do not use.
Step 4: test the physical layer (2 minutes). Swap the cable, move the plug to a different jack, and try a different pair of headphones. If a different transducer works on the same jack, the original hardware failed. If nothing works on any jack, return to the driver layer.
Step 5: rebuild the endpoint stack (10-20 minutes). On Windows, open Device Manager, uninstall the audio device including its driver software, restart, and let Windows reinstall. On macOS, delete the aggregate or virtual device in Audio MIDI Setup, then restart. This fixes endpoint corruption but is wasted effort if steps 1 through 4 were skipped.
Step 6: escalate to hardware only after software is excluded. A codec that is absent from firmware, or a USB audio device that fails on every host, is a hardware fault. Before concluding this, test the device on a second computer; a single-host failure is almost never the hardware.
Set a time budget: if 15 minutes of the top-down sequence has not produced sound, the failure is probably endpoint corruption or a virtual-driver conflict, and step 5 is the next move rather than more troubleshooting.
Limitations: what this diagnostic chain cannot tell you
This sequence assumes the audio hardware is present and powered. It cannot distinguish a dead codec from a firmware-level disable, and it cannot diagnose failures in audio hardware that the operating system never enumerates. It also cannot resolve application-specific bugs, such as a media player that fails to initialize its audio engine on a specific codec, without testing that application in isolation.
This article covers consumer operating systems and standard audio endpoints. It does not address professional audio interfaces with vendor-specific drivers, DAW routing, or networked audio, where the failure modes and recovery procedures differ substantially. Where symptoms persist after the full sequence, the remaining causes are hardware-level or vendor-driver-level, and manufacturer diagnostics are the appropriate next step.
Nothing here is a substitute for manufacturer documentation or professional repair for a device under warranty; opening or modifying hardware can void coverage. If the computer is used for work where silence during a call carries professional cost, keep a known-good USB headset as a fallback endpoint, because a working alternate endpoint converts a potential hour-long outage into a five-second switch.