(L-R) Gary Epstein; Helen Audi; Steve Foster; Tom McAndrew Toronto Section April 26, 2011
This was quite an intensive session with a lot of ground to cover within a limited space of time. Some of the presenters read directly from notes in order to cover all the points they wished to address.
Sy introduced Steve Foster, who is a graduate of Humber College, and has been involved in audio post production for 25 years, editing and mixing sound. He's presently staff re-recording mixer at Technicolor Toronto and also a Genie and Gemini award winner for best overall sound.
Mr. Foster gave an overview of the steps required for bringing various sound elements together to create audio mixes for the Bluray format. Reading from prepared notes, he offered an intensive and concise discussion of the audio requirements for the Bluray format.
Some points: Preparing audio for DVD authoring is not a complicated task. A proper monitoring environment is critical for accurate levels. The main monitor level is often too hot. The dialogue channel often determines the main playback level set by the consumer. Fortunately, there are theatrical specifications which state that pink noise at 0 VU should have a C-weighted SPL value of 85 dB in the centre channel.
The ATSC (Advanced Television Systems Committee) has determined that smaller mixing rooms should have lower playback levels. Basically, monitoring levels mirror those that the average consumer will be playing back at.
Calibrating the room properly is of prime importance. Online resources are available for detailing calibration procedures.
Projects intended for theatrical release require different specs than those for broadcast or DVD/Bluray media - HDTV requirements being similar to Bluray. Dialogue, music and sound effects, more often that not, need to be re-balanced due to the fact many consumers are listening to a player-reduced 2-channel downmix of 5.1 material.
However one major difference is that Bluray/DVD mixes are often full scale, dynamically speaking, often peaking at 0 dBFS. Mr. Foster prefers, in this case, to limit the final output for DVD mixes to -0.1 dBFS.
When queried about this during question period, he stated that this was not an industry standard but a personal preference to prevent clipping indicators being activated at quality control facilities.
He also prefers 24 bit files for mixing as there is a definite increased clarity, an observation that has not escaped his clients.
Another very important point made is that files for video and audio should have the same start times for authoring as there is no embedded time code in the audio files that can be read by the authoring software. And to prevent synchronization issues, he recommends providing a picture file that contains an audio mix as well. The discrete audio files can then be "phased" against that mix to guarantee sync prior to being trimmed and sent off for authoring.
Labeling files with the appropriate channel name included is also of key importance to prevent mistakes in output assignments. Practical experience early on reinforced why labeling is important!
Mr. Foster hoped that with some releases trickling in, there will be a place and desire for uncompressed 96kHz/24 bit Bluray audio much as there is for high definition video.
One final question concerned peak levels again: Mr. Foster was asked if he would adopt a lower full scale setting if he knew in advance that the material would be perceptually coded at lower levels, he replied "no" as he has had no problems currently. He added delivery was always for the highest resolution whenever possible.
Steve thanked the audience.
Next, Sy introduced Gary Epstein, who gave an overview of Dolby's technologies for the BD disc media and its advantages. Mr Epstein, who has been with Dolby Labs for 12 years, is currently product marketing manager for Dolby Media. He is also a member of SMPTE, AES, The Cinema Audio Society, and The Television Academy.
Mr Epstein began with a joke: "Why did the reverb go to the dentist? Because it had a lot of decay"!
He noted this is the first time he and Tom McAndrew from DTS were presenting together in the same forum.
Continuing with this statement: "Lossless audio is the highest quality you can get whether it's from Dolby or DTS", Mr. Epstein went on to discuss the benefits of Dolby's lossless code TrueHD.
The Dolby solution allows multiple playback configurations in one encode. This includes 8-channel (7.1), 6-channel (5.1), or 2-channel with the lower channel counts being a downmix of higher quality channel counts or discrete mixes.
This allows a simple decoder to extract only the data it needs — for example a television with only two speakers will play back the stereo mix. So the contents are defined during the encoding process providing much flexibility.
The trade offs: incorporating 3 discrete mixes involves 16 channels of audio and an increased data rate!
To avoid downmix overload, attenuation is needed. The simplest solution is to attenuate all channels equally to maintain spatial balance, but that would result in quieter mixes for lower channel counts by as much as 7 dB which is 'not a good thing'.To avoid this a process called downmix limiting is applied. When activated in the encoder, it analyzes the audio and applies limiting only when needed. This is represented as meta-data for the consumer's player. The limiter's maximum output is -0.1 dB which is in agreement with Mr. Foster's methods.
The limiting is balanced meaning if a channel element, for instance, needs attenuation, its corresponding opposite receives equal attenuation. An option to attenuate the mix only is available if limiting is not desired.
The system allows encoding both the lossless codecs and the Dolby Digital companion stream simultaneously to conform to Bluray spec. There is no dependency between the Dolby Digital content and the lossless encode, meaning a 5.1 mix is always delivered in the Dolby Digital companion stream.
The encoding has a 'pre-analysis stage' to prevent things like mismatched dialogue levels and digital overloads in downmixes, created in the files before doing an actual encode. Downmixing offers the option to significantly reduce the size of the Dolby TrueHD stream.
Dynamic Range Control (DRC) is included in the TrueHD stream. Each mix has its own set of compression metadata optimized for its presentation. DRC is useful for late-night listening modes, and for sets with smaller speakers incapable of handling the wide, full dynamic range. Survey conducted by Dolby suggests consumers like options like this.
Bitstream editing was discussed. It was developed to improve ease of use when manipulation of Dolby TrueHD bitstreams is needed. Using a device named 'restart headers' allows editing virtually anywhere seamlessly and with audio remaining in sync. This has been integrated into Sonic Scenarist; integration with other programs is currently underway.
Dolby Digital Plus is 7.1 encoding for lossy formats and secondary audio.
A lengthy Q&A ensued where, ultimately, an attendee wanted to know how an intelligent up-mix will be created without synthesizing elements from the discrete mix. Mr. Epstein replied there was no way that could be done to which Tom McAndrew was in agreement. Mr. Epstein also noted that currently there are no consumer delivery formats higher than 7.1 available but public demand would dictate their introduction in the commercial market. PL2X & PL2Z are decoder-only extractions. These are content dependent and personal preference will dictate best choice for movies and stereo audio CD's.
Mr. Epstein gave a brief demonstration Dolby Media Producer software. If the files are labeled properly the Dolby software will place them into the correct playback channel. (This made Steve Foster happy when asked by Gary)! Otherwise, it lets you check audio channels before proceeding. Basically, everything discussed was shown in this software demonstration. Consumer encoding needs to be attenuated 3 dB (because theatrical surrounds are 3 db louder) unless a special remix was made.
A short lossless and Dolby Digital encode job were completed simultaneously. The encoder allows QC'ing the movie once it's done. A short demo was played with a 2-channel downmix.
A quick demonstration of the 'tools' completed the presentation. 'Tools' allows the editing of previously encoded audio.
A question regarding rendering time for 16 channels (7.1 and downmixes) of a 90 minute movie was asked, the lossless codec takes 20% of real time and the Dolby Digital takes 10% and both are done concurrently when done with the network version of the encoder: Mr. Epstein noted there are two versions the encoder. A stand-alone version for smaller facilities and the noted network version for places like Technicolor or Deluxe.
Steve Foster wanted to verify that when downmixing occurred, only the surround data was analyzed and that nothing was done to the LCR, to which Mr. Epstein replied in the affirmative.
Gary thanked the audience.
Sy thanked Technicolor for sponsorship, and Ron Lynch for organizing the meeting. Everyone was invited to have a break and partake of the refreshments.
After the break Sy introduced Tom McAndrew who oversees audio production consumer service. Previously he was Bluray & DVD audio compression manger at Sony Pictures Digital Offering Centre. He also worked in the Washington DC area as a documentary sound designer for Discovery Networks and National Geographic. He's a graduate of Syracuse University and lives in Los Angeles.
He stated he had not been in Toronto since the early '80's. He acknowledged his long, cordial relationship with Gary Epstein.
Mr. McAndrew discussed DTS Audio for Bluray disc, market position, and gave an overview of DTS technologies used on Bluray and backward compatibility, why DTS should be used and the production tools for Bluray and demonstrated Master Audio Suite.
He noted that all the major studios support DTS Master Audio on Bluray disc and expressed pride in having them as partners. A PowerPoint slide indicated DTS' increasing market share for Bluray lossless audio.
He provided an overview of the all DTS technologies in use today starting from DTS Digital Surround (from DVD days) to DTS-HD High resolution Audio to the flagship codec DTS-HD Master Audio. He continually pointed out the backward compatibility feature with each new stage of the technologies.
He also mentioned DTS Express - the low bit-rate codec for secondary audio capable of 5.1 audio. It's also used for BD-Live streaming and downloading.
DTS-HD Master Audio major advantage is that it is the most data & bandwidth efficient lossless audio solution for Bluray, with lowest bit-rates, smallest file size, and backward compatibility meaning that every stream has at its core the digital surround file. It's a single file that any Bluray player can play.
The discussion moved to the DTS-HD Master Audio Suite which encompasses three components: the encoder, the stream player for QC, and the stream tools for file manipulation. He also mentioned Minetonka's SurCode for Apple Final Cut Studio and Logic Studio and their DTS-HD decoder for Apple Quicktime; and Neyrincks SoundCode DTS-HD Stream Player for Pro Tools which functions as a QC device.
The Neural UpMix by DTS plugin for Pro Tools and VST hosts was given attention because of the previous question regarding options for upmixing from 2 channel and 5.1 sources. He noted it as a a great solution for 'budget' upmixing but never intending to take away from the 'dedicated professionals' doing real multichannel mixes. It can provide 5.1 and 7.1 from 2 channel mixes, and 7.1 from 5.1 mixes. He noted the limiter and its recommended ceiling of -0.1 dBFS, again in line with Mr. Forster's criteria. A one-week free demo is available.
Examples of Bluray titles with DTS-HD master audio was displayed on the next slides.
He said using PCM on Bluray audio limits your options and recommended using lossless codecs for music titles for Bluray discs otherwise there would be no room for multichannel programs. Later when demonstrating the software, he noted another reason for using lossless codecs was the ability to customize downmixes.
And being in Toronto, he brought attention to the recent release of Rush's "Moving Pictures" in 5.1 on Bluray DTS-HD Master Audio at 96 kHz and 24 bits.
He provided a demonstration of the DTS-HD Master Audio Suite. This program also provides batch verification and recently added folder-based encoding, useful for encoding an entire season of a television-based series, for example. It too places files in their correct channels if correctly labeled.
Branching was discussed: when theatrical and director's cuts are included on Bluray, the additional scenes are included as extras (to greatly conserve disc space) and data is included for the player to 'assemble' the director's version on the fly. DTS created Seamless Branching Optimization so that the encoding is optimized to offer 'a truly seamless experience' for the listener/viewer.
A question from the audience asked whether this 'storyboarding' wouldn't be done in the authoring program as opposed to the encoder. Mr. McAndrew could not answer, but he added that storyboarding is also done in authoring. The process noted was actually creating a storyboard just for the audio track, and there are additional processes that have to happen for video encoding and authoring to make branching work properly.
Gary Epstein expanded on the answer by stating that the points for branching can also happen in the encoding stage, but the options exist whether to do this during encoding or later for authoring - to which Mr. McAndrew concluded with "What he said!"
The DTS-HD Stream Tools has an add silence features providing a more 'elegant' solution to maintain audio and video sync should the need arise.
The DTS-HD Stream Player simulates every consumer playback situation, as well as providing a very in depth QC environment.
A question asked if phase meters are included, Mr. McAndrew replied metering options are better utilized as outboard equipment as opposed to software. Another question was if metadata was able to be manipulated in the player. The answer was no because that wouldn't affect what was encoded. Problems encountered would need to be fixed in the encoding stage.
Mr. McAndrew then played a demo trailer in 2-channel format.
Further questions concerned processing time. The reply was less than 1/4 real time. With the latest version there is multi-core encoding, so with 12-core towers a 2-hour feature can be done in 20-25 minutes for example, or 12 feature encodes in 20 -25 minutes for Master Audio. Loss-y codecs take even less time. They are done simultaneously. Options are available to enable the number of cores used in the process, freeing up other cores to run other projects or programs.
The automation data is no longer Pro Tools only - any program that exports AAF correctly is essentially valid. Mr. McAndrew did not know whether or not that was .XML data.
Tom thanked the audience.
Sy introduced Helen Audi, the final speaker at this meeting. Ms. Audi has been working in the Bluray/DVD industry for over ten years. In July 2005, she began working for Technicolor in Montreal for two years before being transferred to Technicolor Burbank in Los Angeles where she created DVD's for all major Hollywood studios. Looking for further challenges, she began authoring Bluray and 3D Bluray discs, and moved back to Montreal in 2010 and is currently working for one of the few Bluray authoring facilities in Canada.
Ms. Audi spoke about importing elements, authoring and encoding, QC and Replicating, and the tools and workflows involved in producing a completed Bluray disc. While also reading from notes, she provided much commentary from time to time, and answered questions during her presentation.
There are two types of authoring: HDMV (high definition movie mode) and BD-J (Bluray Disc Java). HDMV is similar to SD (Standard Definition), while BD-J requires Java applications that conform to the BD spec that control the logic and user navigation of the disc.
While HDMV is more popular with smaller studios due to its simplicity and decreased cost, make major studios prefer BD-J due HDMV's limitations. However, one other reason Ms. Audi noted are increasingly shorter turnaround times demanded from clients in her facility, she is able to create a fully authored HDMV disc in 2-3 days.
While an HDMV authored disc has all the functionality of an SD disc, access to local storage of players is blocked and features such as disc tracker are not possible in HDMV. With BD-J you can contral exactly what happens when a user presses a button or any event occurs ie: displaying an on-screen dash when the player is paused.
Ms. Audi displayed a list BD specifications (indicating things as support for video, primary audio, secondary audio, interactive audio, still image, subtitle and menu) while briefly discussing some points.
Steps to authoring a BD disc begin with reviewing the project flowchart. Clients have different preferences as to what comes first for things like logos, warnings and trailers for example. Then the graphic artist designs the menus and Interactive graphics are created with Photoshop and Scenarist Designer PS.
Once menus are fully created, creating movie clips are the next step. Clips are where audio, video and subtitles assets are associated. Clips are similar to timelines in a video editor.
Playlists - where clips are housed - are created next. Playlists are also where pop up menus are integrated and chapter points are created. User ops are also determined here. An example is when a user is blocked from skipping out of warnings; that is the work of someone like Ms. Audi! But a trick to avoid that is to go through the scene selections menu!
A question about whether output formats were HTML or CSS after rebuilding in Photoshop, Ms. Audi replied it was XML. The JavaScript for BD-J is often proprietary. Regarding chapter selections, Ms. Audi wasn't sure if Final Cut had an export option to create timecodes, but a list could be created and that would be imported into Scenarist.
The next step is menu logic and navigation, followed by title logic where playlists are housed in titles. Commands are added to each title which the disc reads when the tiles are entered. These are read till the playlist is called, and when finished with that, it will read the rest of the commands. HDMV authors do not need to be hard core programmers when writing.
MUX'ing is next which is the process of creating the relevant files required to be put on the physical disc. This is also known as a layout. Emulation and QC follow and when all issues are resolved, the layout is ready to be sent out for client approval.
Shipping is next where an image of the disc is sent for client review. The BD-R is sent with the copy protection Indicator off so that it plays in the clients machine. Once approved CPI and AACS is turned on and a final CMF(Cutting Master Format) is created which is sent out to the replication facility.
Ms. Audi answered more questions, one revisiting a point regarding chapter points and backgrounds. Another concerned the physical format sent to replicators - no physical formats are involved, only the CMF is sent. Another one regarding returning a BD disc to a player after having been ejected midplay and re-inserted playing back from its original spot, this is a hardware feature not an authoring point. Regarding the 4 coloured un-named keys on the remote, their functions are dictated by java programming not HDMV.
Helen thanked the audience.
Sy thanked the speakers, sponsors and everyone else for attending,
Final questions about an audio-only BD format: Gary Epstein replied it was slowly moving along and involved changing the specs somewhat, possibly incorporating the 4 coloured un-named keys on the remote to allow 'screen-less' access to the contents. An audience member added that there was an article in the April 2011 journal by Francis Rumsey covering all the formats available for audio-only Bluray Disc.
This brought up another point of how long the plastic disc will be around. Tom McAndrew replied: for new releases quite some time, for catalogue not so much as it's not the same 'gold-rush' for studios as it was for DVD. The audience broke up into smaller groups continuing different discussions.
