Audio Recording in Cameras: KVKK Risks and Steps to Turn It Off on Site

A microphone left on in an IP camera is a separate KVKK risk. Steps to detect and turn off audio on the camera, in the VMS and on the client, and to narrow the scope if needed.
Technical summary
Technology
KVKK Law No. 6698, IP camera, RTSP, ONVIF, VMS, G.711/AAC
Topic
Detecting, turning off and limiting audio recording in camera systems
Level
Intermediate
Prerequisites
  • Camera inventory
  • Camera and VMS admin access
  • A test client with FFmpeg/ffprobe installed
In this article7 sections

When an IP camera’s microphone is left on, conversations in the corridor, phone calls and meeting audio spilling through an open door are written to the recording disk along with the video. Many models have a built-in microphone or an external audio input, and on some, audio encoding comes enabled at factory settings. The installation team looks at framing and focus; nobody notices the audio stream. This article explains for IT and information security teams why audio is a privacy risk separate from video in camera systems, how to detect and turn it off on the camera and in the VMS, and how to narrow the scope when it’s really needed.

Why audio recording is assessed separately

Video records where a person is and what they’re doing. Audio records what they’re saying. A corridor camera’s microphone doesn’t only hear those in the corridor; it also picks up the meeting in the next room, the employee on the phone, a private conversation between two people. With video, you narrow the angle, add a mask, crop the frame. A microphone has no frame.

This difference has direct consequences for the principles in Article 4 of KVKK. Recording the content of conversations on a camera installed for security can’t be defended, in most scenarios, as processing relevant and proportionate to the purpose. In addition, recording non-public conversations between people without their consent can create criminal law risk under Article 133 of the Turkish Penal Code. Legislation texts are followed on mevzuat.gov.tr, and Board decisions on kvkk.gov.tr.

Dimension Video recording Audio recording
Limiting scope Framing, masks, angle Practically none; all sounds in the environment
Whether people notice The camera is visible The microphone often goes unnoticed
Content Location, movement, behavior Conversation content, third parties, sensitive information
Legal risk KVKK principles, privacy notice A criminal law dimension on top of KVKK principles
Technical detection Easy Requires separate checks per stream and device

The rule on site is simple: audio is off by default and turned on as a justified exception. This rule should be in the written policy and verified technically. Off on paper isn’t enough.

Detecting the audio feature on a camera

Keep the audio capability of each camera as a separate field in the inventory; it saves hours at the next audit. Detection is done in three layers:

  1. Hardware: Look for “built-in microphone,” “audio in/out” and “two-way audio” on the datasheet. An external microphone may have been fitted on site to a camera without a built-in one; on models with an audio input, don’t skip the physical check.
  2. Camera configuration: In the web interface, look at the audio menu, the audio encoder in stream profiles (such as G.711, G.726, AAC) and the stream type. With some manufacturers, the stream type is selected as “Video” or “Video&Audio.” A common situation: audio off on the main stream, on in the substream.
  3. Network stream: Verify through an RTSP session what the camera actually sends. Only this check shows whether the setting in the interface has been applied.

The example below lists a camera’s streams with ffprobe from the FFmpeg package. The RTSP path varies by manufacturer; use the path in your own camera’s documentation.

# List the video and audio channels in the stream
ffprobe -v error -rtsp_transport tcp \
  -show_entries stream=index,codec_type,codec_name,sample_rate \
  -of compact "rtsp://audit:[email protected]:554/STREAM_PATH"

# Example output (the second line appears if audio is carried;
# sample_rate is printed only for audio streams)
stream|index=0|codec_name=hevc|codec_type=video
stream|index=1|codec_name=pcm_mulaw|codec_type=audio|sample_rate=8000

On a camera with no audio stream, only the video line comes back. Repeat the check separately for each stream profile: main, sub and, if present, third. The VMS often uses the substream for live viewing and the main stream for recording; checking a single profile and saying “no audio” is half the job. On devices supporting ONVIF, the Media service’s audio source and audio encoder configuration also show audio capability; the onvif.org documents can be used as a reference for profile and service definitions. The roles of stream profiles are explained in the camera multi-stream management guide.

Turning off audio recording on the camera and in the VMS

Turning audio off at a single point isn’t enough. Audio can reappear at any link in the chain from the camera to the client. Turn it off in this order and verify each layer separately:

Layer Action Verification
Camera streams Turn off audio encoding in all profiles; set the stream type to video only No audio line in ffprobe
Camera microphone If supported, disable the microphone/audio input and set the input level to zero Interface screenshot
Edge recording Turn off the audio option for SD card recording No audio channel in a sample clip
VMS / NVR Disable the microphone device or audio channel; remove it from the recording rule No audio in the recording file
Clients Remove audio listening permission in mobile and desktop clients Check with a role-based test account

Some enterprise VMSs list the camera’s microphone as a separate device, and this device can stay enabled independently of the camera. On these platforms, editing the camera’s recording rule doesn’t stop the audio; you need to disable the microphone device itself. On NVR-based systems, look at the stream type in the per-channel recording parameters. That’s the first place to look in the console.

The truly insidious risk is the persistence of the setting. A firmware update, a factory reset, replacing a faulty camera with the same model or the VMS rediscovering the camera can return the audio setting to its default. A changed camera means new audio. That’s why the audio check should be written into the closing step of firmware update and device replacement procedures. If you want to address it together with configuration hardening, the audio setting can be added to the configuration templates under IP camera and NVR cybersecurity hardening.

Narrowing the scope if audio is needed

In some scenarios, audio comes from a legitimate operational need: a two-way intercom at a door or barrier point, a remote monitoring center’s voice warning announcement, operator communication with the control room during an incident. Most of these needs call for audio transmission, not audio recording. If you make the distinction correctly, most of the risk disappears.

  • Speaker output only: For remote voice warnings (e.g. “This area is under surveillance, please move away”), the camera’s audio output is used; the microphone stays off.
  • Two-way audio at the time of an event, without recording: The intercom conversation is held live, and audio recording in the VMS is off.
  • Event-triggered, time-limited recording: If recording is really needed, it’s done only when an alarm or call button is triggered and for a defined duration. No continuous audio recording is done. Event triggering alone doesn’t replace a legal assessment.
  • Point and scope limits: Audio is enabled only on the camera at the relevant point; other cameras of the same model stay off by template.
  • Separate retention and access: A shorter retention period and a narrower access role are defined for recordings with audio; listening and export are written to the audit trail.

Audio analytics (glass break, shouting, gunshot detection) is a separate matter for assessment. These features analyze audio continuously; some implementations produce only event metadata and don’t store raw audio. Verify from the manufacturer’s documentation whether the analytics keeps raw audio, whether processing is done on the camera or the server, and whether audio is added to the event clip. A marketing brochure isn’t evidence. For the general privacy framework of analytics decisions, see the KVKK limits in video analytics article.

Impact on the privacy notice and policy

Audio recording is assessed separately from video. Before enabling the audio feature, the purpose, necessity, legal basis and less intrusive methods should be assessed. A sign, written internal approval or event triggering alone doesn’t meet this assessment. The approach in the Authority’s announcement on workplaces can’t automatically be generalized to apartment buildings, a different field of activity or another country’s rules.

At a point where audio is recorded, a privacy notice and warning sign prepared for video recording aren’t enough. A person who sees the camera expects video recording; they don’t assume their conversation is being recorded. For every point where audio is processed:

  • The warning sign should state clearly that audio is also recorded, and a separate sign should be used at that point.
  • In the privacy notice, audio data should appear as a separate data category, with its purpose, retention period and access scope.
  • In the personal data processing inventory, audio recording should be defined as a separate processing activity.
  • The internal camera policy should contain the rule “audio recording is off by default; enabling it is subject to written approval” and name the approving authority. Written approval is an internal control step; it doesn’t replace the assessment of purpose, necessity and legal basis.

The general structure of layered information is explained in the KVKK camera privacy notice guide; audio recording enters this structure as an additional layer. For policy documents and technical configuration to match each other, the cleanest way is to address both together under KVKK camera monitoring and privacy notice processes. Address access and retention decisions for camera recordings together with the organization’s other data processing activities under KVKK technical and administrative measures consulting.

Audit checklist

Apply this list at least once a year and after every major change. Record findings per camera; “generally off” isn’t a finding.

  • Are the “microphone yes/no,” “audio input yes/no” and “audio status” fields filled in for every camera in the inventory?
  • Have all stream profiles on every camera with audio capability been checked with ffprobe or an equivalent tool?
  • Is any microphone device or audio channel still enabled in the VMS?
  • Is there an audio channel in sample recording files and exported clips?
  • On cameras doing edge recording, does the SD card recording contain audio?
  • For every point where audio recording is on, is there written approval, a justification, a retention period and a separate sign?
  • Which roles have audio listening and export permission; have these roles been reviewed?
  • Has the audio setting been re-verified after the latest firmware updates and device replacements?

If you find an unapproved audio recording point during the check, decide what will happen to existing audio recordings before turning the setting off. These recordings should be brought within the scope of the retention and destruction policy and shouldn’t continue to be kept without justification. Turning off the setting doesn’t erase the past.

Frequently asked questions

How can you tell whether an IP camera has a microphone?

Look at the datasheet first: does it mention a built-in microphone, audio input (audio in) or two-way audio? On site, check the audio menu in the camera’s web interface, the ONVIF audio source information and whether there’s a separate audio stream in the RTSP stream. If the ffprobe output has a line containing codec_type=audio, the stream carries audio.

Is turning off audio recording only in the VMS enough?

No. Even with recording off in the VMS, the camera may keep sending audio in the stream, audio may be heard in the live viewing client, or the camera may record audio to its own SD card. If audio isn’t needed, audio encoding and the microphone are turned off on the camera side, and the microphone device is also disabled in the VMS.

Can audio be recorded in areas such as a checkout or a call point?

It can be considered only with a concrete purpose, a proportionality assessment, separate information and a narrowed scope. Because recording conversations between people can also have a criminal law dimension, get your legal advisor’s written opinion before deciding.

Can audio settings change after a firmware update?

They can. Some updates or a factory reset return stream profiles to defaults. Re-verifying audio status after updates and device replacements should be the closing step of the change procedure.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Video Surveillance

IP cameras, recording and central management infrastructure. We plan blind spots, retention periods and the camera network around the risks of each site.

Knowledge Center

Other articles