Audio Watermarking Audio watermarking for TV program Identification 3Gb/s,(NexTracker HD, SD embedded domain Dolby E to PCM ) with the Synapse DAW88 module decoder with audio shuffler A A product application note COPYRIGHT 2011 AXON DIGITAL DESIGN B.V. COPYRIGHT 2012 ALL RIGHTS RESERVED NO PART OF THIS DOCUMENT MAY BE REPRODUCED IN ANY FORM WITHOUT THE PERMISSION OF AXON DIGITAL DESIGN B.V.
PROCESSING Introduction In a multi-platform, digital broadcasting environment with many new ways of watching TV content, both in real-time and time-shifted, accurate audience measurement has become more difficult. Audience Measurement services must now report more accurately and reliably across a larger number of channels, delivered through an everchanging and diverse mix of broadcast platforms. In response Civolution developed the NexTracker Audience Measurement system. Designed for easy and automated use by media research organizations, NexTracker Audience Measurement makes identification of media listened to and watched by people straightforward even with time-shifted, catch-up TV and podcasts. What is Watermarking? NexTracker Audience Measurement utilizes Civolution s watermarking technology that can be inserted and detected live or from recorded sources, the technology resides in a range of devices that can be easily integrated into broadcast play-outs, within catch-up TV/Podcast services, and into third party panelist audimeters. Civolution s watermarking technology ensures that NexTracker Audience Measurement is imperceptible to consumers though robust to the processing that occurs along a broadcast chain. NexTracker allows instant retrieval of valuable information about the audience and broadcast, such as the content and channel ID, transmission time or catch-up TV/podcast-watched, and if it was listened to by a panelist. Digital watermarking embeds indelible and imperceptible data into a media asset to provide a persistent identity. This data can later be extracted and interpreted to identify, manage or monetize the asset. The audio watermark information survives signal compression and transitions in the broadcast chain. It will travel through the air from the TV speakers so that the Second Screen device can detect it via the microphone input. Digital audio watermarking inserts information into the audio bit-stream or audio file without the information being perceptible to the listener, and without it affecting the audio quality of the original source material. Watermarking is mainly used for the protection of Intellectual Property rights, especially protection against music piracy where it provides information about content ownership and can be used to track the original source of material. Watermarking is increasingly used to identify the source of transmitted material and the time of transmission, information used by audience measuring and second screen synchronization systems. The information that is inserted may be static and simply be the title of the item, name of the channel etc., or it can have dynamic elements such as a timecode indicating the time of transmission or a specific location within a program. Watermarking for Audience measurement Because watermarking embeds information into the audio stream of a program the relationship between this information and the program content is not changed by the route the signal takes from the broadcaster to the viewer even if the program is recorded and played back hours of days later. This makes digital audio watermarking an ideal technology for audience measurement systems, once the channel s ID and the transmission date and time is embedded into the program s audio it will always remain there. Whenever an audience measurement device detects the embedded codes within a program s audio it can use them to determine the program being viewed, this independent of whether the program is viewed live, delayed by the transmission method or replayed from a DVR. page 2
PROCESSING Other uses of Watermarking Digital audio watermarking is based on humanly imperceptible but still audible audio signals, these sounds can be picked up by any device with an in-built or attached microphone such as a mobile phone or tablet PC. By loading a suitable application on to the mobile device it can interpret the watermark data it receives and, depending on the application, display additional information about the program being viewed, allowing the viewer to play-along with a quiz or vote in a talent show. This is known as SyncNow or 2nd Screen. Axon and Civolution also produced a module to support this process (DAW77) and also a module which inserts watermarking for both audience measurement and SyncNow (DAW99). The range of these 2nd Screen or SyncNow applications is only limited by the imagination of the broadcaster. Benefits of the audio watermarking technology Digital Audio Watermarking technology can be quickly and easily integrated within a transmission or post production environment. Because it uses signals embedded within the audio elements of a program it supports all distribution methods including linear playout, Catchup TV, VoD, recordings on personal recorders including PCs and TV broadcast to on mobile devices. The use of audio based identification makes watermarking suitable for fixed meter and portable meter usage as both simply need to be able to hear the program audio in order to detect the media s identity. The watermark detector can even be implemented in a smartphone. Because watermarking identifies the source of the media through uniquely encoding information into the SDI stream adding a separate encoder to each of the output lines feeding terrestrial, satellite, cable or IPTV transmissions would allow audience measurement information to be gathered on a platform by platform basis. The technology is proven to work reliably with over 150 channels being continuously watermarked for audience measurement in Europe and North Africa. The detection time depends on the amount of information embedded into the watermark, but typically, for a fixed meter, one measurement is made every 15 seconds and the accuracy of the TV channel change detection is around 1 second. Audience Measurements Workflow SDI with Embedded Audio SDI with Embedded Audio including Watermarking Compressed Video and Audio Signals DAW88 Presentation System Digital Audio Watermarking Engine Transmission TV Receiver Audio Watermark Reader Timecode or NTP Local Clock BROADCASTER Channel Identifier Catch-up TV / VOD Computer / Handheld PC VIEWER Measurement Data Collection and Processing Audience Report AUDIENCE MEASUREMENT Co. 1. TV channel signal originates from the playout system and is watermarked by the DAW88 module as it is aired with the TV Channel Identifier, the airing date and the time of transmission. 2. The output is transmitted live to the viewer via terrestrial or satellite transmission, or stored and viewed later via a catch-up TV, Video-on-Demand service or personal video recording device. 3. Civolutions NexTracker is hosted on the panelist s audience meter, automatically identifies the Channel ID and Timecode of the program currently watched by the panelist. 4. The audience meter sends periodically (typically overnight) the decoded watermarks to the Audience Measurement company s back-office systems. 5. The back-office system consolidates the data retrieved from all the panelists and generates the audience reports for the broadcasters. page 3
PROCESSING Synapse In order to facilitate the encoding of the Civolution digital audio watermark Axon and Civolution have closely cooperated in the design and development in a range of modules for the Synapse system. The Transmission Time (based on timecode), this accurately identifies the airing time of the original TV program regardless of any timeshift cause by network latency, PVR recording, or Catch-up TV. However broadcasters who operate time-delayed (or +1 channels) commonly insert the watermark information after the delay server as they may wish to separately monitor the size of the audience for the real-time and delayed services. The Content Identifier (for non-linear content) which individually identifies the content or the program stream e.g. Catch-up TV. The DAW88 is a 16-channel AES watermark encoder module, utilizing both local AES connections and the Synapse Quad-Speed Audio Bus. Depending on the implementation, the watermark will carry one or more of the following piece of information: A Channel Identifier which uniquely identifies the TV channel A Network Identifier which can be used to identify the transport mode (Satellite, DTT, DSL ) Example 1: DAW88 block schematic DAW88 [QUAD SPEED BUS] ACTIVE BYPASS AES/EBU IN 1/2 AES/EBU IN 3/4 AES/EBU IN 5/6 AES/EBU IN 7/8 SRC SRC UP TO 8 CHANNELS OF NexTracker SRC 40x8 INPUT MUX SRC AES/EBU OUT 1/2 AES/EBU OUT 3/4 AES/EBU OUT 5/6 AES/EBU OUT 7/8 HIGH POWER DSP SUB MODULE REAL TIME CLOCK WITH BATTERY BACK-UP 32 CHANNEL TIME CODE (LTC) 2 CHANNEL 2 CHANNEL 2 CHANNEL 2 CHANNEL ADD-ON EMBED MULTIPLEXING OPTIONAL RE8 X 2-CHANNELS INTO 32 INSERTING OF PROCESSED COMP GPI (2 IN/2 OUT) 1 µp 32CH IN FROM NEXT ADD-ON 32CH 32CH OUT OUT TO MASTER MUX TO NEXT ADD-ON REFERENCE INPUTS 2 32CH IN FROM MASTER PLL QUAD SPEED MULTIPLEXING BUS RACK CONTROLLER INTERNAL SYNAPSE BUS Time Reference For an audience measurement system to be effective it needs to be able to accurately identify when a viewer is watching a particular channel. To enable this the encoding device is required to insert a precise timestamp into the watermark and therefore the watermarking engine requires a source of accurate time. The DAW88 has a local connection for Linear Timecode (LTC) which is commonly distributed within broadcast centers, it can also use Network Time Protocol (NTP) which is frequently distributed across professional IT networks. NTP is also readily available from a number of sources on the Internet. NTP distribution to modules is be supported in the current range of rack controllers (ERCxx). However the previous Synapse rack controllers (RRCxx and RRSxx) are not able to distribute NTP directly to the modules in a frame. In this instance NTP can be supported by running an instance of Cortex, Axon s control and monitoring application. Details of how to freely obtain Cortex and deploy it for this use are given in the DAW88 Quickstart Guide available from the Axon website www.axon.tv. If neither LTC nor NTP is available the DAW88 has an internal, battery backed-up, clock system which can be used to generate the time to be used by the watermarking engines. page 4
PROCESSING DAW88 configuration Licensing The DAW88 requires two licenses; one from Civolution and another from the Audience Measurement organization. The Civolution license specifies the number of channels that can be watermarked and also the application type. The Audience Measurement license specifies a unique TC channel identification what is this as well as the maximum allowed drift of the system time. Generally the Civolution license will already be installed on the module when it is delivered, the Audience Measurement license will be sent from the Audience Measurement company. If the users wished to change the DAW88 s application type or the number of channels that can be processed and upgraded license can be obtained from Civolution and loaded by the user. Example 2: Cortex screen showing DAW88 time source management Current time and date stored in the module s clock time the clock system will manually synchronize to if selected by Reference Time Source LED showing if source for Reference Time is available and valid Source to be used (if available) to automatically sync to the on-board clock Controls for manual synchronization of the onboard clock to LTC, NTP or manually entered time and date Example 3: Cortex screen showing DAW88 license Usage Number of watermarking engines that the Civolution license allows Number of watermarking engines that have been allocated page 5
PROCESSING Audio Routing The DAW88 module has three audio routing configurations; the default setting configures the module to utilize the local AES connections as 4 inputs and 4 outputs. In this mode 4 of the module s 8 Watermark Engines can be accessed. The other two configurations use a combination of the local AES connections and the Synapse Quad-Speed Add-On Audio bus to provide routing to/from all 8 Watermark Engines. The three routing configurations are shown in Example 4. In addition when the module s Synchronization Mode is set to MasterCard (the module s reference is derived from clock signals originating from the MasterCard via the Add-On bus) the Quad-Speed Add-On Audio bus can be used to route audio between a MasterCard, such as a HDK100, and the DAW88. The configuration of this routing is shown in Example 4, Example 5 and Example 6. The DAW88 can use AES1, the MasterCard or the reference signals distributed across the frame as its local reference. Routing signals to or from the Quad-Speed Add-On Audio bus is only enabled when the module is using MasterCard reference. Example 4: Cortex screens showing DAW88 audio routing configurations Configuration 1 Configuration 2 Configuration 3 The DAW88 is configured to use its local AES connection for both input and output. In this mode only 4 of the available watermarking engines can be used. The DAW88 is configured so that all the local AES connections are inputs to the module. This provides local audio inputs to all 8 watermarking engines, the output is returned to the MasterCard via the QuadSpeed Add-On bus. The DAW88 is configured so that all the local AES connections are outputs from the module. This provides local audio outputs from all 8 watermarking engines, the output is also returned to the MasterCard via the QuadSpeed Add-On bus. Note; audio from the MasterCard can also be used as the source for some or all of the watermarking engines if required. In this mode the input to the watermarking engines comes solely from the MasterCard. If the Synchronization Mode was changed to MasterCard the other 4 engines could use audio connections provided by the Quad-Speed Add-On bus. Note that the LEDs indicate that a signal is present on the named input. page 6
PROCESSING Example 5: Stereo watermarking with audio from local connections Operating mode of the watermarking engine: Stereo Source of audio: Local inputs 1/2 Status and format of the audio input to the watermarking engine Example 6: Watermarking a 5.1 surround sound mix with audio routed from Mastercard Operating mode of the watermarking engines showing the 5.1 surround mix is being processed through 3 engines. Source of audio: ADD-ON bus slots 1-6 Status and format of the audio input to the watermarking engines: the level of channel7 (Left Surround) is clipping Note: only Lf, Rf and C have watermarks embedded, the other channels are passed through the system unwatermarked Watermark engine status showing that 2 engines are active (for Lf/Rf and for C) in a 5.1 surround sound Example 6 shows the audio routing configuration for encoders 3/4, 5/6 and 7/8; the left-hand boxes show that the audio is routed from the Quad-Speed Audio Bus (Master channels 1 to 6), the centre boxes show the audio format is PCM and that the status of five of the tracks is OK whilst the level of the audio on channel 7 (Surround Left) is too high. The right-hand boxes show which track pairs of the 5.1 audio are being processed by each engine and which tracks watermarks are being embedded. The first engine (for channels 3 and 4) is active and fully utilized, the second (5/6) is active for one channel, and the third (7/8) is not active. Example 7: Quad Speed ADD-ON bus audio output routing The Quad Speed ADD ON bus has 32 audio slots. The DAW88 is configured to insert the output of all of its watermarking engines in the first 16 slots. Here slots 17 to 24 have been configured to pass the audio from the local inputs directly on to the MasterCard Example 7 shows the output audio routing from the DAW77 to the MasterCard using the Quad Speed ADD-ON bus. The bus is capable of conveying 16-pairs of audio tracks, the configuration shown here places the output of the 8 watermark embedders onto the first 8-pairs and then also routes the audio from the 4 local AES connectors onto the next 4 Quad Speed Bus pairs. page 7
PROCESSING Associated Synapse modules There are several modules within the Synapse In recognition of the widespread use of embedded range which utilize Civolution watermarking audio in broadcast facilities the same watermarking technologies for audience research, NexTracker technology is also available on a range of modules (2nd Screen) or a combination of both. The DAW with video connections covering SDI, HD-SDI and range presents the watermarking engines with AES 3G. audio connections. Synapse modules with Civolution watermarking technology: Audience measurement Second Screen Both technologies (NexTracker + NexTracker ) AES DAW88 DAW77 DAW99 SD-SDI SAW100 SAW010 SAW110 HD-SDI HAW100 HAW010 HAW110 3Gb/s SDI GAW100 GAW010 GAW110 Example 8: Embedded watermark embedder block schematic GAW/HAW/SAW010/100/110 3Gb/s, HD, SD IN 1 [QUAD SPEED BUS] 3Gb/s, HD, SD OUT 1 2x1 Eq ADJUSTABLE OFFSET 0-1000ms AUTO MUX 3Gb/s, HD, SD IN 2 Eq 16 CHANNEL EMBEDDER 3Gb/s, HD, SD OUT 2 AUTO PHASER 16 CHANNEL DE-EMBEDDER 56X16 MUX 16 Ch 16 DE-EMBED 32 QUAD SPEED 8X WM 4 4 4 4 Ch Ch Ch Ch 16 CHANNEL PHASE + 24Ch ACTIVE BY-PASS TIME CODE (LTC) 48X8 MUX COMP 16 DE-EMBED 32 QUAD SPEED PLL NexTracker (100/110) SyncNow (010/110) HIGH POWER DSP SUB MODULE REAL TIME CLOCK WITH BATTERY BACK-UP GPI/O (2x out 1x in) µp TO ADD-ON 32CH IN FROM ADD-ON 1 32 Ch TDM MUX/DE-MUX 32CH OUT REFERENCE INPUTS 2 UP TO 8 CHANNELS OF QUAD SPEED MULTIPLEXING BUS Rack controller INTERNAL SYNAPSE BUS Example 8 shows the block diagram of the embedded Watermarking modules. The module can be specified with SD, HD or 3G inputs and outputs and modules can be field upgraded. The modules have the same Watermarking embedding capabilities as the separate audio units and additional Watermark embedding engines can be added by connecting further, separate audio modules, via the Quad Speed Audio ADD-ON bus. page 8