PMD Studio is an application for authoring professional metadata. It provides a simple user interface for configuring audio beds, objects and presentations. The authored metadata may be saved as an XML file in either Serialized ADM or PMD formats or streamed using a professional sound card or a Nvidia/Mellanox Connect-X Smart NIC. The latter requires
Serial ADM is defined jointly in ITU-R B.S. 2076 and 2125. Streaming Serial ADM output is supported over AES3 and SMPTE ST 337 which is defined in SMPTE ST 2116.
PMD (Professional Metadata) is defined in SMPTE RDD 49. A copy of this document is available to the public on the Dolby Professional Developer Portal. PMD output is supported over AES3 and SMPTE ST 337 which is defined in SMPTE ST 2109.
PMD Studio is supported on Mac and Linux only. Professional Sound Card support is provided for both operating systems but
Tested on Ubuntu 18.04 and 20.04 for Linux and OSX Mojave. There is a known issue with OSX Catalina that the menu-bar is unresponsive. The simple workaroud for the issue to take focus away to another application and then return focus. The menu-bar will now respond normally.
The application supports real-time input of pre-rendered audio and combined output of rendered audio and metadata.
To support real-time I/O, one of the following hardware options is required
- A Professional multichannel sound card with ALSA (Linux) or Core Audio Drivers (Mac)
- An NVIDIA ConnectX SmartNIC supporting the Rivermax SDK.
The presence of the drivers can be checked using the -dl switch or checking the device lists in the settings panel. The clock settings must be sane such that input and output intefaces are using the same clock. The application has been tested on the folowing hardware:
- RME HDSP MADI card on Ubuntu
- Digigram LX-IP with on Ubuntu
- BlackMagic Decklink 4K Extreme on OSX
PMD Studio can be simply evoked from the command line without any options. Some basic options are provided at the command line but most of these can be overidden with settings in the application.
- -fb File-based mode (overrides streaming options)
- -f Filename of file to load on launch
- -l Input and output latency in seconds
- -device Device specific option
- -buf Buffer size in samples
This mode is forced when no streaming audio interfaces can be found. The -fb switch allows the user to select file-based mode even when audio interfaces are available. In file-based mode streaming audio and metadata outputs are not available. XML files can be opened, edit and saved. Certain metadata options are only available in file-based mode. These include divergence for objects and extra configuration options for beds.
-device list provides more information about the available hardware devices
These settings can be used to alter the latency PMD Studio requests from the sound card and the buffer size that is used for data transfer between PMD Studio and the driver. Nominal values are chosen by default. To minimize latency smaller values may be selected. Depending on the performance of the system decreasing latency may introduce audible artifacts. On older or slower systems it may be necessary to increase latency to eliminate artifacts. A modern multi-core system should have little difficulty with the default settings.
When launched the application puts the audio streaming interface into pass-through. This can be useful for checking basic connectivity. The pass-through configuration is muted once changes are made on the user-interface.
The user interface is split up into four sections:
- Audio Beds
- Audio Objects
- Audio Presentations
- Outputs
Each section has an 'add' button that creates a new entry to be populated by the user. The beds and objects sections should be configured to reflect the pre-rendered audio being presented on the streaming audio input.
Beds are audio elements composed of at least two channels. They follow a normal channel configuration such as 5.1. In streaming mode stereo, 5.1 and 5.1.4 beds are supported. The channel ordering for a 5.1 bed is L, R, C, LFE, Ls, Rs. For 5.1.4 the channel ordering is L, R, C, LFE, Ls, Rs, TFL, TFR, TBL, TBR where TFL is top front left (T = Top, F = Front, B = Back, L = Left, R = Right). The field start refers to the first physical channel of the bed on the input audio multiplex. Multiple beds may refer to the same physical channels allowing different gain settings for the different bed entries.
Audio objects work in a similar manner to beds but only represent a mono audio element. This position of the audio element can be set using Cartesian coordinate settings.
The 'En' checkbox enables or disables the bed or object.
The Presentations panel defines how the beds and objects are combined to form presentations. Presentations are also know as presets or preselections.
One bed is allowed per presentation and is selected using the drop-down menu. The 'lang' specifies the language used by the audio in the presentation. Typically this will be dictated by the language of included dialog objects, but beds can be premixed with dialog as well, for example in a CM+AD (Complete Main plus Audio Description) presentation. NLang or Name Language field specifies the language used by the presentation description in the name field. Selection is by ISO language code. Three methods exist for the setting of the name languagage. The first of two automated methods is for the name language field to follow the presentation language used by the audio. For example, using the following name strings:
- English
- Francais
- Deutsch
Alternatively the name language can be made to fixed to a single setting defined in settings panel (follow preset nlang). For example, if the preset NLang is set to English then the name strings could be:
- English
- French
- German
The last method for the setting of the language is unlocked and allows a custom nlang setting. In this mode the nlang setting is enabled on the main panel and can be set to any language code independent of the audio presentation language setting.
The user interface can be reset using the 'New' option under the file menu. PMD or ADM files can be opened or saved using the file menu. The settings option opens the settings panel. When the application is shut down via the quit option in this menu the last known settings are saved in 'pmd_studio.cfg' in the current directory. This file is loaded by default when PMD Studio is invoked to restore the last known settings.
When using a professional sound card an Audio Device Configuration menu option will be available. This allows selection of the input and output audio devices and the number of channels to be used.
When using an NVIDIA Rivermax enabled NIC, a stream setting menu option is available. This allows selection of the input and output streams. Input streams are discovered via multicast DNS and RTSP (Ravenna), SAP (Dante) or NMOS IS-04. The IP address and port of the NMOS registry can be explicitly defined or left blank to be discovered via multicast DNS.
The startup procedure is as follows:
On first invocation open the stream settings panel and select the correct interface, PTP Clock Domain, annnoucement methods and add at least one output stream. Apply the settings and close the panel. Reopen the panel and select the correct input stream. Add any necessary output streams.
Only a single input stream can be selected but multiple output streams can be defined. For each outputs stream the name of the stream is defined as well as the output codec to be used and the channel allocation. Channels are automatically allocated in order based on the number of channels in each stream. Available codecs are SMPTE ST 2110-30/AES 67 with either 16 or 24 bits, SMPTE ST 2110-31/AM824 or SMPTE ST 2110-41. The former should not be used for metadata outputs as SMPTE ST 2110-30 / AES67 is defined to only carry linear PCM. Output streams are advertised on the network using Ravenna, Dante or NMOS methods.
Once the stream settings are applied the IP subsystem will be restarted with the new settings. A progress bar may be displayed when searching for a Dante stream. This process can take up to 30 seconds. If the input stream cannot be acquired, an error message will be displayed.
The update option creates three file representations of the audio metadata:
- Serial ADM XML in sadm.xml
- PMD XML in pmd.xml
- PMD KLV binary format in pmd.klv
A script called 'update' in the current directory, if it exists, is then executed. This function can be used to drive some method for sending the metadata to a down-stream device. For example, a common use of this functionality is to perform an HTTP Post of the XML to an encoder device.
Two types of outputs are available: Audio and Metadata outputs. Audio outputs are always rendered as per one of the presentation definitions. The presentation to be rendered is selected using the down-drop menu. The output channel configuration can be selected between stereo, 5.1 and 5.1.4. The start channel in the output audio multiplex is defined. The last channel used by that output is calculated and displayed. When the 'En' or enable checkbox is selected the audio output will be live. Multiple audio outputs are possible by using the 'Add Audio Output Button'.
Each metadata output be be configured to generate Serial ADM or PMD output as selected by the format drop-down menu. The rate at which metadata is generated can be selected using the frame-rate drop menu. Common broadcast rates are available. The Mode selection allows the use of either frame-mode or subframe-mode as defined in SMPTE ST 337. Frame mode places the metadata across two consecutive channels. Subframe places all the metadata in a single mono channel. The mapping of metadata and audio outputs is achieved by a matching of the channels specified on the output controls to the streams defined in the streams settings panel. For SMPTE ST 2110-41 metadata streams a mono channel is sufficient. When a metadata output is enabled PMD Studio enters live mode. In this state the metadata configuration is fixed except for gain and positional settings.
PMD Studio provides a simple Ember+ consumer (client) that allows an Ember+ provider (server) on the same network to control a restricted set of metadata values. The server IP address and port are configured from the settings panel under the file menu. Connection to the server or console is initiated using the Connect option under "Console" on the menu-bar.
NOTE: The module is currently hard-coded for provider software that is not publicly available, and so not intended for use by the average user.
OSX Catalina requires focus to be moved from and to application to enable menu-bar. Audio elements and outputs can be added to the user interface but cannot be removed. Unused elements and outputs should just be disabled.
Custom Ports are not supported.
Certain features relating to NMOS IS-04 and IS-05 are not implemented. Basic patching of connections has been tested with Riedel NMOS Explorer.
NVIDIA, Rivermax and ConnectX are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and/or other countries. BlackMagic and Decklink are trademarks and/or registered trademarks of Blackmagic Design Pty Ltd. Digigram is a registered trademark of Digigram S.A.