Triton Digital's ad insertion technology relies on metadata cues points from publishers to determine when to replace or insert commercial content. For the best listener experience, cue points need to be as close as possible to the perceivable audio transition.
Also see:
When cue points are perfectly aligned, the experience is seamless: The listener perceives inserted ads as if they were part of the original broadcast content, without noticeable interruptions and without hearing a portion of the underlying original content.
This alignment is important at both the beginning of an ad pod and on the return-to-content marker, which we call the ENDBREAK cue point in Station Manager’s RAS Adapter. ENDBREAK determines where to reconnect the original content after the replacement content has been played.
Factors that Affect Cue Point Alignment
There is no guarantee that metadata is perfectly timed to the first frame of the played-out audio. Many factors, such as crossfade position and the audio infrastructure, can affect the alignment between metadata and audio. In relation to its audio, cue point alignment can be described in a few ways:
Late in audio. Cue points arrive after its corresponding audio point in the audio.
Ahead of the audio. Cue points arrive after the corresponding audio.
Constant, where the cue point difference is consistently late or ahead during playback. A small variability of 1 to 30 ms is barely perceivable and thus should be considered as constant.
Variable, where different crossfades define the transitions as different times.
Altered by audio delays, profanity filter delays, or other tools affecting the path of metadata.
Cue Point Misalignment from Station to Station
There are many reasons why cue point misalignment can be different from station to station, including these:
Audio fed through an audio delay system.
Audio routed through an audio processor before, or on, the streaming computer.
Automation system configuration set to delay the metadata in the configuration.
Sending the data over the internet instead of direct peer-to-peer internally.
Data routed through multiple networks, firewalls, or packet scanners.
Metadata being fed through multiple third-party systems.
The computer on which the automation system runs is not up-to-date so it processes the output of the metadata slowly or at variable rate.
Profanity delay that reconstructs its buffer without properly realigning the metadata feed. These cases are difficult to correct and might require help from your profanity delay provider.
Cue Point Offset
You can correct some kinds of misalignment with a cue point offset, which specifies a positive or negative difference between audio and cue point metadata.
A cue point offset can correct misalignment in the following case:
Incoming cue point misalignment is constant
Misalignment is late or ahead by a range within +5000 to -15000 milliseconds
A cue point offset cannot effectively correct misalignment when cue points vary or are altered beyond +5000 to -15000 milliseconds. However, it is possible that the cause for misalignment can be isolated and removed from the audio chain, thus making cue point offset fall within the addressable cases. Note that even within this range, the farther away from zero offset the more difficult it might be to obtain a stable solution.
For example: a system constantly sends metadata 500 ms ahead of the beginning of an audio segment. After entering a cue point offset of -500 ms in Station Manager, the outgoing cue point is then aligned to 0 ms.
Suggested Areas to Investigate
0 ms – 450 ms late or ahead: This is typical and should be regarded as normal. Use the Station Manager cue point correction to adjust. You likely will not want to change anything else, since other systems are possibly already taking this offset into account, such as RDS encoders and HD Radio Exporters.
450 ms – 800 ms: This is usually found in systems that read the metadata from a filesystem to produce the cue point to send. You can reduce this by avoiding writing files over network drives, as they can be a source of variability. There might be faster ways, for example XML over TCP, to send the data that your automation supports. You should look into this if you have too much variability on the cue point offsets. Otherwise, this is well handled by the Station Manager cue point alignment setting.
800 ms – 5000 ms: This is still within the range that Station Manager can attempt to correct. This often happens when using certain audio processing settings. Be careful when you change audio processing settings as your delay might also change. The typical 1800 ms is often a sign that you are feeding your encoder POST an existing on-air processor, which is generally not recommended as it is better to have a specific audio stream processing chain.
Over 5000 ms: Cue point alignment has limited effectiveness in this situation. Instead, start by looking at your audio chain to reduce any audio delays and other likely culprits mentioned above. If the offset remains constant and can be reduced to under 5000 ms, then the methods described above can be used to finish the alignment.
Maintaining Alignment
Maintaining cue point alignment requires some basic maintenance. Here are some tips that will help in maintaining alignment for as long as possible.
Synchronize the clocks on all of the computers that are involved in the delivery of content, including your automation system, using Network Time Protocol (NTP). Achieve this by making sure you use the same time server for your clock synchronization. In Windows 10, for example, use the Set the time automatically feature in Date and Time settings. This prevents having one machine drift away slightly every day.
Measure and verify the alignment by listening to the stream at least once every three months. A scheduled reboot at night might help realign, or you can add or subtract a few milliseconds from the Cue Point Alignment Setting in Station Manager.