Maintaining a Consistent Audio Path

How Belo deals with 5.1 surround and stereo audio for DTV


(click thumbnail)Jerry Paonessa, engineering supervisor at KENS, with the MultiMerge installation. Audio for the DTV transmission path has proved challenging for many stations. One issue is how to handle switching between 5.1 surround and stereo audio.

For many of the 19 stations in the Belo television group, it used to be the traffic department's responsibility to keep track of which programs and commercials were in 5.1 and which were in stereo.

"We used to put in a secondary event in the automation system to trigger a GPI to switch the Dolby encoder between the different modes," said Reed Wilson, technology manager for broadcast media at the Belo Television Group.

But switching audio this way led to viewer complaints that the center channel would jump to the side speakers.

To address this, stations and groups are tending to provide a constant surround sound signal for the DTV main channel, upmixing stereo to surround where necessary, and setting the Audio Coding Mode (acmod) on the Dolby Digital encoder to a fixed or 3/2L.

Belo decided to follow this concept with 11 of its stations installing the Neural Audio MultiMerge. Wilson said that most of the stations use the Harris FlexiCoder ATSC MPEG-2 encoder with the MasterPlus HD master control processor module which has two HD inputs and one SD input with upconverter and logo capabilities.

Since the MasterPlus switches video with embedded audio, the output is run through an external audio de-embedder to extract the AES audio. AES is then fed through a TFT 940A for EAS, and then into the MultiMerge.

"The MultiMerge auto senses whether the input is stereo or 5.1 and always creates a 5.1 mix," Wilson said. "No longer do we need to GPI anything to switch between 2/0 or 3/2 mode."


According to Dave Casey, product line manager for Neural Audio, "MultiMerge uses a patent-pending process of detection to determine if content is 2.0 or 5.1. When MultiMerge detects a stereo input, it automatically upconverts the audio to 5.1. There really is a .1 to the 5.1 as MultiMerge creates an LFE channel. MultiMerge uses the Neural Surround algorithm for up conversion and blends to UpMix mode instead of abruptly cutting over. Neural Surround is 100 percent backwards-compatible to L/R matrix encoded content. The upmix results in a stable and natural-sounding image without the traditional upmix issues broadcasters are familiar with such as dialog leakage or unstable surrounds."

The output from MultiMerge then feeds a Dolby DP569 AC-3 encoder, with acmod set to 3/2. This in turn feeds the FlexiCoder which provides SMPTE310 to the STL. (The FlexiCoder does not contain an internal AC-3 encoder.)

The FlexiCoder also takes in the main channel HD video as well as SD subchannels. The SD subchannels, like NBC's WeatherPlus or stations' local weather channels, typically have stereo audio which is encoded in a separate Dolby DP569.

Another Belo station, NBC affiliate KGW in Portland, Ore., chose the Linear Acoustic OCTiMAX 5.1 audio processor as its upmixer.

"We didn't have the correct facilities to deal with the upconverted signal," said Tim Kerr, technical systems manager at KGW. "So we generated a station capital project to do it before the corporate level Neural Audio project."

The impetus for getting something on the air sooner was the 2006 Torino Winter Olympics. The station was also in the process of adding an HD master control switcher, the Grass Valley M-2100, and the new Grass Valley K2 shared storage servers for commercial play-out, all under automation control by a Sundance Titan. The project was completed about two weeks before the Olympics began.

Inputs to the HD M-2100 include an IRD with the NBC HD feed with 5.1 audio, a second IRD with the NBC HD feed but with stereo audio, the station's upconverted SD feed with stereo audio, and the outputs from the K2 servers. While the K2 servers provide 16 channels of embedded audio, they provide two AES outputs, Kerr said. The audio inputs to the M-2100 are AES.

The audio output of the M-2100 is a set of three AES pairs which feed distribution and patching, and for the first two pairs, a backup switcher. The first pair also feeds a TFT 940A for EAS.


These three AES feeds are input to the Linear Acoustic OCTiMAX 5.1. The first AES feed is also copied into a fourth AES input on the OCTiMAX. (When a stereo source is switched on the M-2100 its output appears on the first AES pair.)

When the M-2100 makes a switch to a stereo source, the tally output acts like a GPI to trigger the OCTiMAX to switch to the fourth AES input (stereo) and then upmix the audio to 5.1. Whether the input to the OCTiMAX is stereo or 5.1, the output is always 5.1, and this is what feeds the Dolby DP569 encoder.

"The upmixer in OCTiMAX 5.1 is called upMAX, a proprietary algorithm developed by Linear Acoustic, " said Tim Carroll, founder and president of Linear Acoustic. "We generate a true 5.1-channel output. The LFE channel is created by default, but can be shut off if desired, and some customers choose to do this for music mixing. upMAX provides realistic surround sound and is completely downmix and matrix decoder compatible, meaning that stereo and Pro Logic or PLII listeners also get the best audio possible."

As for metadata, Kerr said that the metadata from the NBC IRD is connected to the OCTiMAX input. The Dolby DP569 then receives metadata from the OCTiMAX, which passes through NBC metadata, but changes the acmod value to 3/2L when it upmixes.

OCTiMAX also provides audio level processing. "Level processing is automatically applied to both the 5.1 and the 2-channel programming to match network, local, and commercial programming," Carroll said. "The loudness control follows suit in the background and does not normally require additional operator intervention, although GPI control can be used to make processing changes if desired."

KGW installed a Wohler multichannel level meter panel in the master control room to confirm that the feed on the network HD 5.1 IRD really provides six audio channels. If by chance stereo is sent instead, the master control operator can take the HD with stereo IRD feed to be upconverted.

"We want to turn on the surround light on all the receivers," Kerr said.

The audio processing through the OCTiMAX and Dolby Digital encoder produces audio delays. "We adjust for lip-sync at the FlexiCoder by setting the video delay," Kerr said.

The audio for the upconverted SD feed must be de-embedded and then put through an AES delay before feeding the HD master control switcher M-2100.

"End users like it," Wilson said. "It allows simulated 5.1 and the dialog is always in the center channel."