Audio Volume Targeting

Audio Volume Targeting is now available in SpringServe if you have turned on the Audio Volume Targeting beta feature. With this feature, clients now have the ability to exclude demand sources from eligible supply based on a specific attribute of the media file in the VAST response, namely the average audio volume of a media file. As an example, if a supply source wants to ensure that only media files with a volume less than -20 db average volume play on their supply, this can be achieved by using audio volume targeting and the behavior of this feature is listed in detail below. 

Audio Volume Targeting in the UI

Once this feature has been enabled for your account by your Account Manager, you will find a new section on the targeting tab of applicable Supply and Demand Tags called Audio Volume Targeting. This is a targeting sub-section in the MediaFiles targeting section This feature works on all Managed and Connected Supply & Demand Tags with the following exceptions:

  • It is not eligible on any Managed vpaid enabled Supply tags

  • On Connected Supply, not eligible for Mobile-Web, In-App and Desktop Environments.

This feature is highlighted in the screenshot below with the default selections for this targeting parameter.

Details of Audio Volume Targeting in the UI

The default selection for Audio Volume Targeting is All, and when Custom is selected, there are two options available to users, namely Mean Volume and a checkbox to Block Unknown Volume. 

For the screenshot above, based on the selections made we will show media files that have an average volume less than and including -22 decibels. Additionally, because the "Block Unknown Durations" Checkbox is selected, we will also exclude all media files from vast responses that we haven’t as yet scanned and thus haven't been able to detect the average volume for.

Audio Volume Targeting in the Ad Server

It is important to note a few things about how this is handled in SpringServe's AdServer. The first thing to note is that because both Supply and Demand Tags are eligible for Media File Targeting, if Media File Targeting is applied to both Supply and Demand Tags, then this targeting is applied on each applicable tag in each applicable link of the call chain. Secondly, because of the way the VAST spec works, the Audio Volume Targeting attribute is applied a bit differently to the VAST responses. As a rule of thumb:

  • If a VAST Response from a Demand Source fails the Audio Volume Targeting checks, then we remove individual Media files from the VAST response before sending it to the video player.

It is critical to note that volume settings are not displayed in the VAST Response anywhere and these are settings that SpringServe’s proprietary Creative Scanning tool has automatically detected for this media file.

Audio Volume Targeting Reporting & VAST Error Codes

Additionally, if users want to use reporting to troubleshoot errors in reporting, you can run a Vast Error Code Report with the appropriate supply and demand tags as filters and find error metrics specific to Audio Volume Targeting, namely:

  • 1646: Targeting Block for Audio Volume.

SpringServe’s Creative Scanning Tool

For this targeting feature to be eligible for your account, its is important to note that we need to add the creatives seen in the VAST responses on the supply of your account to our creative scanning tool. Once we start scanning these creatives, we are able to detect certain attributes of the creative like the average audio volume and pass this information to the AdServer for targeting purposes. It is critical to note that our tool can’t scan vpaids and thus if these are seen in your VAST responses, these will be treated as unknown by us.

How we detect volume

To detect the volume, our creative scanning tool scans all the media files seen in the VAST responses on the supply of your account. We initially download the media files offline and then to detect the volume, we use certain programmatic machine learning libraries. These libraries run an average of different sound settings on multiple plays, and then comes up with a statistical distribution to define what the average across the multiple plays was and then we take the average across each of these individual video plays and assign that value to the creative.

The Decibel Scale

As shown in the screenshots shared earlier in this page, this targeting attribute uses a decibel scale. Decibels are a unit used to measure sound levels. Decibels are not an absolute measure of sound but rather a comparison with a reference value. Decibels are similar to the Temperature Scale where 0 Degree Celsius is set as the temperature at which water freezes over to ice. Similarly, 0 dB is considered the aural threshold of the human ear when dealing with professional sound equipment and video imaging. In this context, 0 dB is the maximum value that a creative can have and -200 dB is the minimum value that a creative can have.

The Decibel Scale in the UI

Given the context above, the maximum values that our creative scanning tool detects for the average volume of a creative is 0 dB and the minimum value for this same measure is -200 dB and this is also the range of the scale in the UI. In our case, a volume of 0 dBs is indicative of a peak signal level supported by the hardware encoding standards used by digital and analog audio hardware systems. What this means for the end user is that because of the video encoding standards used in online advertising, 0 dBs is the loudest sound a human ear can hear and anything lesser than -200 dBs is completely silent from a human hearing perspective.

For further reading, and more technical details of the decibel scale, please refer to this primer described here. You’ll see that as an example, as an audio standard, Netflix sets the limit for peak volume at -2 dBs for the videos that they play on their streaming service. Note that this should not be used as a reference for setting audio volume thresholds in the UI, but more for the purposes of highlighting how the human ear perceives settings on a decibel scale in a real world example.