# Best practices

<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->

These recommendations help you ship a reliable audio experience with the Video SDK for Windows.

## Select the right audio devices

On Windows, the SDK captures and plays back through the system audio devices exposed by Windows (WASAPI). On machines with more than one device, don't assume the default is correct.

-   Enumerate devices with `getMicList` and `getSpeakerList` and let the user choose, or select a known device by ID. For details, see [Select a microphone or speaker](/docs/video-sdk/windows/audio/advanced/#select-a-microphone-or-speaker).
-   Confirm at least one capture device is present before calling `startAudio`. A server with no audio hardware has no devices to select.

## Enable audio with a clear action

If your app has a user interface, tie the call to `startAudio` to a visible, user-initiated action like a "Join audio" or microphone button. This makes it easy for users to understand when they're connected to session audio.

## Show mute and talking status

-   Render each user's mute state from their `ZoomVideoSDKAudioStatus` so participants can see at a glance who is muted. Implement [`onUserAudioStatusChanged`](/docs/video-sdk/windows/audio/audio-events/#get-notified-when-a-users-audio-status-has-changed) and re-render when it fires.
-   Drive a talking indicator from the [`onUserActiveAudioChanged`](/docs/video-sdk/windows/audio/audio-events/#get-notified-when-a-users-active-audio-changes) callback together with the `isTalking` field. This gives users confirmation that their mic is capturing their voice.

## Server and headless deployments

Some Video SDK for Windows apps run without a user present, for example a bot or recorder deployed to a Windows Server host.

-   If the host has no physical audio hardware, work through [raw audio](/docs/video-sdk/windows/raw-data/) instead of system devices.
-   To inject your own audio into the session without a capture device, provide a virtual audio source rather than calling `startAudio` against hardware. For details, see [Send raw data](/docs/video-sdk/windows/raw-data/send-raw-data/).
-   To consume session audio for transcription or recording, [subscribe to raw audio](/docs/video-sdk/windows/audio/advanced/#subscribe-to-raw-audio-data) rather than routing it to a speaker.

## Sound options

Leave [original sound](/docs/video-sdk/windows/audio/sound-options/) disabled unless your use case specifically needs unprocessed audio. The SDK's default audio processing (echo cancellation, noise suppression) produces a better experience for general voice communication. Enable original sound for music, broadcasting, or other content where the SDK's processing would degrade the source.

## Handle host-ask-unmute on the client

When the host calls `unMuteAudio` on a remote user, the target device receives [`onHostAskUnmute`](/docs/video-sdk/windows/audio/host-controls/#ask-a-single-user-to-unmute). Always show the user a confirmation prompt; never silently turn on their microphone. Calling `unMuteAudio` on the local user without the user's explicit acknowledgment violates the user's expectation of mic privacy.