to generation GEP100 - HEP100 3Gb/s, HD, SD embedded domain Dolby E to PCM SoundField based up-mixing with the module decoder with audio shuffler A A product application note Quad speed Upgradable to 3Gb/s Embedded Metadata S2020 COPYRIGHT 2011 AXON DIGITAL DESIGN B.V. COPYRIGHT 2012 ALL RIGHTS RESERVED NO PART OF THIS DOCUMENT MAY BE REPRODUCED ANY FORM WITHOUT THE PERMISSION OF AXON DIGITAL DESIGN B.V.
Introduction Although there is widespread adoption of High Definition (HD) video many broadcasters have found that not all of their infrastructure can support the surround sound audio often associated with it and they are therefore forced to retain, at least in part, sound stereo PCM operations. Examples of this are production studios that have only two-channel audio equipment installed, editing and server systems may not support sufficient audio channels to accommodate a surround sound mix, or the Media Asset Management system may not provide workflows for mixes. To overcome these situations a broadcaster may need to convert any material delivered with a surround sound mix to stereo, pass the new stereo signal through their system and then expand the stereo back to surround sound for transmission. Many broadcasters find that even new material may be recorded in stereo and when this is transmitted alongside surround sound mixed programming the viewers listening experience dramatically changes between programs. A similar situation exists when commercial (spot) material that is mixed in stereo being played in the ad-breaks within a surround sound mixed program or film. A possible solution to this issue would be to detect when the audio material being transmitted is stereo and use an up-mixer to produce a mix in realtime. Synapse module Axon have collaborated with Soundfield to integrate the up and down mixing from their X-1 processor into the, standard form-factor, Synapse module. The module accommodates two complete processing systems, each normally occupying a 1-RU box. Each Soundfield processing engine can be configured to provide up-mixing, down-mixing or both simultaneously. Software and hardware up-mixing tools have been generally available for some time, but most of these devices utilize effects such as reverb and phaseshifting to create the additional channels required for the surround sound outputs. Instead, Soundfield use a unique algorithm to perform detailed, real-time analysis, of the stereo source from the direct sound. Mono sources (dialogue, commentary, etc) are extracted from the stereo signal and fed to the centre channel and a stable frontal stereo image is maintained by keeping the direct sound sources at the front, with the extracted natural ambience being fed to the rear surround channels. Down-mixing traditionally uses a simple matrixed conversion of the surround sound channels, however Soundfield employ similar analytical techniques to those used in up-mixing to determine the relationship between audio channels. The user has the option of an adaptive equalization to ensure an open and natural stereo image is produced as well as the ability to control the surround downmix levels of the direct and ambient sources individually to avoid ambience build up while retaining the direct sources. Diagram 1: Soundfield processing engine flow diagram Upmix and downmix with auto detect and X-fade Auto detect and X-fade Upmix Downmix Auto detect and X-fade page 2
Diagram 2: block schematic [QUAD SPEED AUDIO BUS] Soundfield processor AES/EBU 1/2 AES/EBU 3/ AES/EBU 5/6 AES/EBU 7/8 ORIGAL ORIGAL 2.0 UP AES/EBU OUT 1/2 AES/EBU OUT 3/ AES/EBU OUT 5/6 AES/EBU OUT 7/8 HANDLG AES/EBU 9/10 AES/EBU 11/12 AES/EBU 13/1 AES/EBU 15/16 8x16 ORIGAL 2.0 AES/EBU OUT 9/10 AES/EBU OUT 11/12 ORIGAL 16x16 Soundfield processor UP 32 CHANNEL AES/EBU OUT 13/1 AES/EBU OUT 15/16 HANDLG OPTIONAL RESERTG OF PROCESSED AUDIO OUT TO NEXT FROM MASTER EMBED MULTIPLEXG 8 X 2-CHANNELS TO 32 µp FROM NEXT PROC TO MASTER RACK CONTROLLER TERNAL SYNAPSE BUS The module can be configured as a stand-alone unit where all audio connections are made via its local, unbalanced, AES connections, or used in combination with other Synapse modules where additional audio connections can be made using the Quad-Speed Audio Bus (QSB). Further information about the QSB is given in Audio Connections later in this AppNote. Workflows There are a wide range of applications where audio may need to be converted from surround sound to stereo or stereo to. Below are three examples of typical situations where up or down mixing is required. The internal architecture of the broadcast facility is only able to handle stereo audio and therefore all incoming material has to be converted (downmixed) to stereo. Note if required automatic loudness processing should applied after any subsequent up-mix stage to ensure that the final signal is compliant following all of the processing stages. In this example the Synapse DLA3 module is providing loudness control across all of the 6 channels making up the surround sound mix. Once the stereo audio has passed through all of the necessary processing it can be up-mixed to a surround signal. The module has two completely separate processing engines each capable of both up and down mixing and both are also able to detect the format of the incoming audio and switch the processing in or out of circuit as required. Therefore if the is placed in the incoming signal path it can automatically produce a down-mixed stereo signal from a surround mix, or allow an incoming stereo signal to pass through unprocessed. A second processing engine, which could be on the same module, can then be used to upmix the stereo material to provide a surround output. Example 1: Conversion of for use in a stereo environment and re-conversion to for transmission Surround to Down-Mix Equipment DLA3 to Up-Mix Loudness Processing Surround page 3
Example 2: Automatic up-mixing of mixed material to match surround sound transmission to Up-Mix OR Automatic Source Selection (with crossfade) Many broadcasters find themselves in the situation where some of their material is stereo (commercials, news, legacy programming) and some is mixed in surround sound (movies, live sport), ideally they want to produce a constant listening environment for their viewers and therefore need to produce a mix from the stereo source material. Loudness Processing Surround The module can detect the format of the incoming audio and automatically produce an up-mixed from the stereo source material if required. Example 3: Automatic up and down mixing of mixed source material to provide and stereo transmission feeds to Up-Mix Automatic Source Selection (with crossfade) Loudness Processing Surround Automatic Source Selection (with crossfade) Loudness Processing OR to Down-Mix It is common that broadcasters have to continue to produce feeds for their legacy standard definition services in parallel with newer HD channels. One method of achieving this is to use a single HD transmission system and down convert the output signal for use on the SD service. SD services tend to have only stereo audio associated with them and so as well as producing a surround sound mix for the HD channel the broadcaster is required to produce a stereo mix, from the same material, for the SD channel. The module s processing engines can utilize the same input audio and apply the necessary processing so that both a and stereo output are maintained no matter what format the incoming material is in. page
Installation and configuration of the module Audio connections The module has 8 local AES connecters on the rear module, in the most basic installation discrete audio can be connected to the module using these connections. video MasterCard, audio can not only be passed from MasterCard to Add-On but also audio from one Add-On card can be subsequently processed by another. The module has two independent processing engines, each capable of simultaneous up-mixing and down-mixing, therefore to fully utilize the processing power of the module another 8 AES connectors would, in theory, be required. Most audio within a broadcast facility is carried as embedded data in the SDI stream, by physically positioning the module next to a video module with a de-embedding function, the embedded audio becomes available as sources to the processing engines. Axon designed the Synapse modular system with the inherent ability to provide audio connections between adjacent modules, originally this Add-On bus allowed 16-channels of audio to be carried either to or from the module. A later development, the Quad-Speed Audio Bus now allows 32-audio channels to be carried to and from each module. Up to 3 Add-On modules can be associated with each A number of Synapse module also allow processed audio to be re-embedded, so by placing the next to a HRB100, audio can be de-embedded, up and/or down mixed and re-embedded into the SDI stream on just two cards with no external wiring, see Diagram 3 below. Diagram 3: Synapse and HRB100 modules used to extract stereo audio from an SDI stream, up-mix to and re-insert in the original SDI stream GRB100 - HRB100 3Gb/s, HD, SD 1 3Gb/s, HD, SD 2 [QUAD SPEED AUDIO BUS] 3Gb/s, HD, SD OUT 1 Eq 2x1 ADJUSTABLE OFFSET 0-500ms AUTO Eq 16 CHANNEL EMBEDDER S2020 SERT 3Gb/s, HD, SD OUT 2 META DATA Program SDI stream input to re-embedder module DE-EMBEDDER SDI 1 16 DE-EMBEDDER SDI 2 6X16 16 DE-EMBED 1 16 DE-EMBED 2 16 32 QUAD SPEED S2020 READER Audio de-embedded from input video passed to via Quad Speed Add-on bus HANDLG 32 TDM PLL TDM DE- OUT GPI/O µp TO 2 FROM REFERENCE S 1 /OUT audio from re-embedded by the HRB100 module QUAD SPEED MULTIPLEXG AUDIO BUS Rack controller TERNAL SYNAPSE BUS Up-mixed audio sent to re-embedder via Quad Speed Add-on bus [QUAD SPEED AUDIO BUS] Soundfield processor AES/EBU 1/2 AES/EBU 3/ AES/EBU 5/6 AES/EBU 7/8 ORIGAL ORIGAL 2.0 UP AES/EBU OUT 1/2 AES/EBU OUT 3/ AES/EBU OUT 5/6 AES/EBU OUT 7/8 HANDLG AES/EBU 9/10 AES/EBU 11/12 AES/EBU 13/1 AES/EBU 15/16 8x16 16x16 Soundfield processor ORIGAL 2.0 UP 32 CHANNEL AES/EBU OUT 9/10 AES/EBU OUT 11/12 ORIGAL AES/EBU OUT 13/1 AES/EBU OUT 15/16 HANDLG EMBED MULTIPLEXG 8 X 2-CHANNELS TO 32 OUT FROM NEXT OPTIONAL RESERTG OF PROCESSED AUDIO TO MASTER PROC TO NEXT FROM MASTER 2 input channels to up-mixer selected from de-embedded audio ΜP RACK CONTROLLER TERNAL SYNAPSE BUS page 5
Diagram : Synapse, DLA and HRB100 modules used to extract audio from an SDI stream, loudness process the audio and down-mix to stereo before re-embedding the audio GRB100 - HRB100 3Gb/s, HD, SD 1 3Gb/s, HD, SD 2 [QUAD SPEED AUDIO BUS] 3Gb/s, HD, SD OUT 1 Eq 2x1 ADJUSTABLE OFFSET 0-500ms AUTO Eq 16 CHANNEL EMBEDDER S2020 SERT 3Gb/s, HD, SD OUT 2 META DATA DE-EMBEDDER SDI 1 16 DE-EMBEDDER SDI 2 6X16 16 DE-EMBED 1 16 DE-EMBED 2 16 32 QUAD SPEED S2020 READER Audio de-embedded from input video passed to DLA Loudness processor via Quad Speed Add-on bus HANDLG 32 TDM PLL TDM DE- OUT GPI/O µp TO 2 FROM REFERENCE S 1 /OUT Audio originating from is reembeddded into the SDI stream QUAD SPEED MULTIPLEXG AUDIO BUS Rack controller TERNAL SYNAPSE BUS Up-mixed audio sent to re-embedder via Quad Speed Add-on bus DLA3 AES/EBU 1/2 LfRf & & AES/EBU OUT 1/2 LfRf AES/EBU 3/ C/LFE AES/EBU 7/8 & & AES/EBU OUT 3/ C/LFE LOUDNESS CONTROL AES/EBU 5/6 LsRs AEROMAX PARAM. EQ OUTPUT AES/EBU OUT 5/6 LsRs AEROMAX 2.0 PAR EQ AES/EBU OUT 7/8 Lt/Rt LOUDNESS CONTROL OUTPUT DE-EMBED OR 8 CHANNELS OUT OF 32 PLL 1 2 OUT TO NEXT REFERENCE S FROM MASTER MULTIPLEXG 8 CHANNELS TO 32 FROM NEXT GPI ΜP OUT TO MASTER OPTIONAL RESERTG OF PROCESSED AUDIO QUAD SPEED MULTIPLEXG AUDIO BUS RACK CONTROLLER TERNAL SYNAPSE BUS Audio from HRB100 is passed through the DLA Loudness processor. The processed audio is placed back on the Quad Speed Add-on bus audio from passed through the DLA unprocessed [QUAD SPEED AUDIO BUS] Soundfield processor AES/EBU 1/2 AES/EBU 3/ AES/EBU 5/6 AES/EBU 7/8 ORIGAL ORIGAL 2.0 UP AES/EBU OUT 1/2 AES/EBU OUT 3/ AES/EBU OUT 5/6 AES/EBU OUT 7/8 HANDLG AES/EBU 9/10 AES/EBU 11/12 AES/EBU 13/1 AES/EBU 15/16 8x16 ORIGAL 2.0 AES/EBU OUT 9/10 AES/EBU OUT 11/12 UP 32 CHANNEL AES/EBU OUT 13/1 Down-mixed audio is sent to the HRB100 via the Quad Speed Add-on bus AES/EBU OUT 15/16 HANDLG OPTIONAL RESERTG OF PROCESSED AUDIO EMBED MULTIPLEXG 8 X 2-CHANNELS TO 32 OUT TO MASTER FROM NEXT PROC TO NEXT FROM MASTER Processed audio from DLA is used as the input to the down-mixer 16x16 Soundfield processor ORIGAL ΜP RACK CONTROLLER TERNAL SYNAPSE BUS page 6
configuration The following configuration examples are based on the user interface provided by the Cortex GUI. Cortex is Axon s control and monitoring application and can be freely downloaded from the Axon website, follow the link below and then select Axon Cortex control and monitoring software: http://www.axon.tv/en/support/downloads/ software. Processing Engines The processing engines can each be configured to operate in one of six modes: Simple Upmix: Up-mixed and Thru Simple Downmix: Thru and Down-mixed Upmix with Auto Detect and XFade: Thru and Thru if present otherwise Up-Mixed Downmix with Auto Detect and XFade: Thru and Thru if present otherwise DownMixed Upmix and Downmix with Auto Detect and XFade: Thru if present otherwise Up-Mixed and Thru if present otherwise Down-Mixed Standalone Up- and Downmix: Up-mixed and Down-mixed These modes are assigned to a Preset on the Engine Mode Parameters tab (see Example below) and then the Preset is recalled as part of the overall engine configuration. Example : Cortex engine mode parameter tab Two of the operating modes use an auto-detect system to determine when the processing engine should cross-fade to the derived from the up-mixed stereo material, this cross-fade is implemented when the audio level in the user configurable channels (C, Ls/Rs or C/Ls/Rs) drops below -100dBfs (see Example 5 below). Two other modes auto-detect the presence of stereo audio, if this is missing the engine will cross fade to downmixed audio from the source material. page 7
Example 5: Configuration of channels used to trigger auto detection The has two processing engines each being independently configurable, Example 6 shows the Cortex screen for Engine 1. In this example all of the audio inputs are derived from the Master(card) via the Quad Speed Add-On bus. This example shows it is possible to use the same source for both and elements of the processing engine. Therefore if stereo material is supplied with Left and Right on Tracks 1 & 2 and also material uses Tracks 1 to 6 the engine is able to detect the presence of audio on the user configured additional tracks used in the mix and route the original to the output. However if the source material only consists of stereo on Tracks 1 & 2 this will be up-mixed and automatically routed to the output of the engine. In addition to audio routing the operating mode, the user settings for the up-mixer and down-mixer and the source and some user settings for the metadata are determined in other Preset screens. Which ones of the these Presets are to be active is determined by 6 select boxes within the Engine s configuration screen. Example 6: Processing engine configuration screen Audio routing and status. Master = Audio from Video Card Local = Audio from local connectors on the module. In this example all audio is from the Master(card) connection to the via the Quad Speed Add-On bus. The operating mode of the Processing Engine. The adjustable parameters of the Upmix and Downmix and the source and content of the surround metadata are determined by various Presets which are separately configured on other screens page 8
The Processing Engines provide user configurable parameters to tailor the up and down mixer to suit the type of programming they are being used with, Example 7, shows the controls available for the up-mixer. The controls shown can be used to change the amount of audio from the stereo source being sent to the various surround sound channels. The Direct Sound control increases or decreases the level of direct sound by + or - 6dB. The Front Ambient Sound control increases or decreases the level of ambient sound in the front Left and Right channels by + or - 6dB. The Rear Ambient Sound control increases or decreases the level of ambient sound in the rear Left and Right surround channels by + or - 6dB. The Width control enables the Direct Sound in the original stereo image to be made wider - at its most extreme the Direct Sound will be in the rear channels. The Centre Divergence control takes the centre channel and diverges it to front Left and Right by the amount set on the control. When no divergence is set all mono material will appear exclusively in the centre channel (hard centre). When divergence is set to full, all mono material will be sent to the front Left and Right channels equally and the centre channel will be muted (phantom centre). Any other divergence setting will create a mix between hard and phantom centre. Enabling the Matrix Mode allows the upmix algorithm to use any matrix encoded information that might be present in the. Example 7: Processing engine up-mix parameter Because the module is Preset based up to 8 different configurations can be stored and recalled, together with the engine s operating mode and metadata set, by the user. Selection can be either by recalling a Preset on the Cortex screen, an automation or other remote control system which controls the module using the Axon Control Protocol (ACP) via an Ethernet connection on the Synapse frame, or by triggering the directly using the local GPI connections. Audio Routing The has both local audio connections and can also send audio to and receive audio from the Quad Speed Add-On (QSB) bus. Its local connections are, by default, AES inputs and AES outputs, however this can be re-configured by the user to provide 8 AES inputs or 8 AES outputs, in which case the remainder of the connections would be via the QSB. Input audio routing is configured on each Engine s configurations screen (see Example 6) and all local and QSB inputs are available for each engine input. The cards local outputs are arranged on a 1-to-1 basis with the engine s outputs, i.e. the Left and Right channels of the section of engine 1 are output from AES/EBU OUT 1/2 (if the module is configured to have 8 outputs).in -in -out mode output connections are only available for Engine 2. The can also send audio back to a MasterCard using the QSB. The QSB uses time division multiplex (TDM) technology and has 16 2-channel slots available. The Cortex Quad Speed Bus tab provides the interface to configure which audio is placed in which slot. Example 8 shows the outputs from Engine 1 (Eng1 to 3 are the surround mix and Eng the stereo mix) are routed to slots 1/2, 3/, 5/6 and 7/8, and the outputs from Engine 2 (Eng5 to 8) are routed to slots 17/18, 19/20, 21/22 and 23/2. page 9
It is also possible to route the outputs from the processing engines into slots on the input QSB, this would be needed if the audio required further processing, for instance loudness control, and was to pass via the bus to the input of another, Slave card, module, in a similar way to the audio passing from the DLA to the in Diagram. Example 8: Quad Speed Add-on bus configuration The outputs from Processing Engine 1 (Eng1-3 are the surround mix and Eng the stereo mix) are routed to the first slots on the Output QSB Example 9: Mastercard (HRB100) Quad Speed Add-on bus configuration The source of audio for the embedder is set to Add-On slots 01-16 and the first channels from that group have been selected Audio Gain, Phase and Delay adjustments for every embedder channel are provided The second embedder has its audio selected to originate from channels 5-8 selected from QSB slots 01-16 The Mastercard has embedders and these can be assigned to any of the audio groups. Be default Embedder A is assigned to Group 1 etc. For the audio from the Processing Engines to be re-embedded into the SDI stream the Mastercard also requires some configuration, Example 9 shows a typical Cortex configuration screen for the embedder section of a Mastercard, in this example it is a HRB100. The Mastercard has assignable embedders each can use audio that is present on the card (from local connectors or de-embedded from video if the card is also a de-embedder) or from the Quad Speed Add-On bus. To use the audio from an Add- On card such as the the audio source has to be selected to come from the Add-On bus. In Example 9 the Add-On selection is further sub divided between slots 01-16 and 17-32. The audio for each channel within the audio Group also needs selection, in Example 8 audio from Engine 1 was placed in the first slots (annels 1-8), to insert this into the first two audio Groups embedders A and B on the Mastercard select Addon01/16 for all channels and then the individual Add-On channels are also assigned to channels within the audio Groups. page 10
(SOUNDFIELD ) Contact information The Netherlands (Headquarters) Asia Russia Office: Axon Digital Design B.V. Lange Wagenstraat 55 5126 BB Gilze, The Netherlands Phone: +31 161 850 50 Fax: +31 161 850 99 Email: info@axon.tv United Kingdom Office: Axon Digital Design Ltd. 1 Forest Court, Oaklands Park Wokingham Berkshire, RG1 2FD United Kingdom Phone: + 118 97 080 Fax: + 118 978 328 Email: info-uk@axon.tv USA Office: Axon USA Inc. 267 Fifth Avenue, Suite 908 New York, NY 10016 USA Phone: +1 212 683 672 Toll free: +1 866 757 9890 Fax: +1 212 265 602 Email: info-us@axon.tv Office: Axon Digital Design B.V. Room 80, 2nd Department, No. 1 Building Beijing Image, No. 115 Fucheng Road Haidian District, 100036 Beijing, ina Phone: +86 10 881 199 Fax: +86 10 881 199 Email: info-cn@axon.tv Southeast Asia Office: Axon Digital Design B.V. 6 Tagore Drive, #02-13 Tagore Industrial Building Singapore 787623 Phone: +65 66 523 016 Fax: +65 65 530 220 Email: info-sg@axon.tv Middle East Office: Axon Digital Design B.V. P.O. Box 10718 Dubai, United Arab Emirates Phone: +971 50 651 1982 Email: axonmideast@eim.ae Office: Russian Branch Axon Stremyannyi Pereulok 38 Floor 6, Room 9 115093 Moscow The Russian Federation Phone: +7 985 217 3218 Email: info-ru@axon.tv Contact your local representative at www.axon.tv/contact www.axon.tv page 11