SourceAudio Detect
FAQ

Fingerprinting-based solutions require features extracted from the detected audio track to be matched against a database of stored features to determine the identity of the track. These solutions rely on the probability of the match between a piece of audio and what is stored in the database. With SA Detect, a unique, robust, and inaudible digital identity (watermark) is redundantly embedded directly into the audio itself.

In theory, yes, but not in reality. Different audio tracks can have similar features (e.g., use the same drum/instrument loops) thus making fingerprinting solutions prone to false positive identification, i.e., confusing one track with another and reporting the wrong track. With SA Detect, there is never any guesswork or doubt about track identity because a truly unique code is woven into the fabric of each audio file that travels with it indefinitely. That code will only reference the true identity given to it by its owner or authorized licensor.

Until now, comprehensive monitoring of production music has been difficult since tracks are often used for very short durations and 
in dense “dirty” mixes. These realities present significant identification challenges for prevailing fingerprinting technologies. The watermark that powers SA Detect (Digimarc) was designed for reliable yet speedy detection in noisy conditions, and is therefore well suited to the challenges of monitoring production music. This reliability leads to accurate reports by addressing the uncertainty associated with missed detects and false matches present in other fingerprinting-based solutions.

Fingerprinting can never distinguish between two or more re-titled versions of a track because the fingerprint is the same for both. With SA Detect, multiple re-titled versions of the same recording can have their own unique, embedded identifiers, which can reference as many non-exclusive, re-titling distributors as are authorized to license the track.

The Digimarc technology was designed for robust ambient detection in noisy real-world environments. The robustness has been validated through rigorous tests involving transcoding across multiple file types, “lossy” compression, and low-fi speaker output/microphone capture in high noise conditions. This same robust design easily ensures the reliability required for monitoring production music on broadcast and cable television.

SA Detect can make a positive identification in as little as 0.2 seconds, unlike fingerprinting-based solutions, which typically require at least 5 seconds of audio to begin making accurate detections.

The watermark in its simplest explanation is a digital code that indisputably references identifying information such as track title, ownership and/or licensing authority. Since these attributes often change in today’s dynamic music licensing marketplace, they are stored in a database that can be easily updated. This way the watermark can be re-directed as needed to the right copyright owner or payee.

Digimarc uses advanced psychoacoustic methods that model the nuances in human hearing and these methods are used to place the watermark inaudibly into the audio track. Double blind tests with music industry experts – including sound engineers, recording engineers, and producers – confirm that the watermark doesn’t distort the audio.

The watermark encoder supports sample rates of 48k and below.  For the time being, audio with higher sample rates cannot be watermarked.  You will receive an error notification with options to either re-upload your files at an acceptable sample rate or for SourceAudio to down-convert your files for you.

Only music copyright owners (music labels and publishers) or non-exclusive (retitling) distributors should watermark their owned or represented tracks.

No. Sub-publisher attribution will be obtained from the reference database and does not require separate watermarks.

Using SourceAudio to host, deliver and manage the metadata on your tracks is the simplest and most efficient way to facilitate use of SA Detect, but you can still take advantage without being a full SourceAudio subscriber. A self-serve, limited use account will allow you to watermark your files and receive detection reports through your own secure dashboard.

You can apply to be part of our free Beta trial at www.sourceaudio.com/detect, or contact your SourceAudio representative and they’ll get you started!

Once you opt-in all audio files uploaded to your account will be watermarked.

Encoding is automatic and seamlessly integrated into the process of publishing your audio files and metadata on your SourceAudio site. A progress bar and automated messaging will alert you on the status of encoding as it occurs.

For large catalogs, please contact support@sourceaudio.com to set up credentials for downloading your watermarked files using an FTP client such as FileZilla. For all other catalogs, our File Delivery Preset is the way to go. First, go to the "File Delivery" tab in your admin panel and click the button on the right to add a new recipient. In the "Preset" drop-down, select the "Watermarked File Download" preset. Then simply put the recipients email address or FTP information in so we know where to deliver the files. Lastly, select the content you want to deliver and you're good to go!

Once fully deployed, our monitoring network will include the 9 major broadcast networks, over 160 cable channels, and over 100 local broadcast stations. The full list of monitored stations and markets will be published very soon -- stay tuned.

Your reports are available through a convenient, interactive dashboard with useful drill down capabilities, intuitive sort and view functionality, and options for quickly and easily exporting reports into common formats.

Reports will tell you the date, time and name of the program (i.e., episode, film, or sports event) your music occurred in, as well as the broadcast, cable or local channel that aired it. If your music was used within a commercial within a program, the name of that commercial will be identified as well. Commercials include brand advertisements, station promos, and theatrical film trailers.

Track detection and program level data are updated in real time. For example, if your music is detected today in an episode of a television series, that detection will appear in your report as it occurs along with the corresponding show title and episode title.

Commercial level data, including names of ads, promos, and movie trailers, is updated weekly. If the aforementioned detection actually occurred within a commercial spot within that TV episode, you will see this additional identifying information in your reports when you check back a week later.

Watermarking inherently involves adding new information to your audio files. As such, newly watermarked music must be distributed to program producers, synched in their programming, and that programming aired before detections can occur.