Dolby TrueHD with Advanced 96k Upsampling: First Impressions

Discussion in 'Blu-ray and UHD' started by Adam Gregorich, May 20, 2012.

  1. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    [​IMG]

    Lossless is lossless. So how can lossless audio get any better? Dolby has found a way and the best part is you likely don’t need to upgrade anything in your system to hear the benefit.

    Home Theater Forum was one of a select group of sites recently invited to Dolby to demo some new products, one of which was their new Advanced 96k Upsampling. When I saw the 48k/96k comparison demo on the agenda I thought for sure that I wasn’t going to be able to hear a difference. The only “golden ears” I have are attached to a hat I got at Disneyland!

    The Demos
    A group of 8 of us filed into a mid-sized room (approx. 15x25) along with several Dolby employees including Craig Eggers, Director of Content Creation and Playback, Poppy Crum, Senior Staff Scientist, and Rhonda Wilson, Senior Member Technical Staff. The demo was conducted by Senior Application Support Engineer James Spezialy using the following system configuration:


    http://www.hometheaterforum.com/image/id/860983/width/600/height/402

    The demo consisted of short (20 second to one minute) TrueHD encoded clips:

    • Joe Satriani: Live in Montreal,

    • The Lost Bladesman,

    • Batman Returns,

    • Kung Fu Panda,

    • Flowers of War

    • San Francisco Symphony Orchestra playing Also sprach Zarathustra, Op. 30.


    They were first played in with Advanced 96k TrueHD and then immediately played in 48k TrueHD. We were instructed to focus on the clarity and naturalness, longer “ring out” to reverb and ambience on the high frequency sounds, consistent audio quality during frequency decay and better definition between instruments on the concert demos and on the more natural sounding dialogue.


    During the first round of demos, I sat in the front row center and was amazed at the overall difference in sound. On some clips it was subtle, on others --- dramatic. It was interesting that different people in the room seemed to be drawn to different differences between the two encodes. The Joe Satriani clip was of the “count off”. Several attendees focused on the wood on wood of the drumsticks during the count off and other high frequency sounds like the cymbals. While I could hear some subtle differences with those sounds, for me there was a dramatic improvement with parts of the midrange with specific notes. The clip from the Lost Bladesman was a fight sequence in a narrow outdoor corridor. The high frequency sword hits sounded smoother and the mid-range “swooshes” that the fighters made as they were dodging each other were crisper. The differences were subtle, but there were definitely audible. We heard two clips from Batman Returns: Flying into the skyscraper in Hong Kong and chasing the Joker in Gotham City on the Batcycle. There was a high frequency countdown timer and a lot of glass breaking in the first scene all of which weren’t as harsh on the 96k. The effect on the second clip was more subtle to me but there were some improvements heard when a trash can was knocked over. The scene from Kung Fu Panda was the hardest for me to hear a difference in. Some attendees heard some improvements in the voice of Ian McShane (Tai Lung), or in some of the effects sounds, but the differences I personally heard in this clip were minimal. We didn’t have video for Flowers of War or the San Francisco Symphony. On Flowers the overall clip was cleaner. I had a better understanding of the room environment where the scene was occurring. The Symphony playing Also sprach Zarathustra was another clip with a huge difference with increased clarity and a much clearer separation of the individual instruments. For the second round of demos, we switched seats and I moved to the side of the back row. The differences I heard from that seat were more subtle, but still noticeable.

    Why Dolby’s Advanced 96k?
    There are several advantages to Dolby’s Advanced 96k Upsampling. DACS work better at higher sampling rates, with fewer artifacts. Dolby’s Advanced 96k Upsampling has a big advantage over just upsampling a 48k signal to 96k as they have licensed Meridian’s apodizing filter. Standard filters used in the analog to digital conversion process leave distortions or artifacts called pre and postringing around the converted sounds. Preringing is interpreted as distortion by the brain and creates a harsh tone, but postringing is ignored as it’s masked by the actual sound you are hearing. Meridian’s filter eliminates the preringing, moving all the artifacts to the postring where it is ignored by your brain:


    [​IMG]

    Conclusion
    Based on my limited exposure to Advanced 96k Upsampling and conversations with the Dolby the impact of using it seems to vary based on the material, the person listening and their environment. There is no downside to this for the end user. From the content creator’s perspective, the 96k encode does take up more space on the disc, but Dolby has some ways that are transparent to the listener to minimize this during the encoding. An Advanced 96K Upsampling encodes takes about 2.5 times longer than the 48K version to encode, but Dolby is working aggressively to cut down the encoding time. For content creators currently using Dolby TrueHD, the next version of Dolby Media Producer will add the ability to do the Advanced 96k Upsampling encoding simply by clicking a button.

    In some cases, the improvement of Advanced 96k Upsampling was subtle, in others it was more pronounced, but regardless --- it is an improvement, and one that just about all consumers can benefit from with no additional cost to consumers or to the content creators. Home Theater Forum sees only upsides to Advanced 96k Upsampling and urges all content creators using TrueHD to quickly adopt it, and for our readers to give it a listen.

    More Information:
     
  2. moovtune

    moovtune Stunt Coordinator

    Joined:
    May 3, 2010
    Messages:
    150
    Likes Received:
    55
    Location:
    Los Angeles
    Real Name:
    James
    I'd be curious to know how the samples you heard were derived. Did they take 48K examples and upconvert them using the new 96K Upsampling software? Is that the comparison you heard or did they have actual 96K recordings encoded with the new software and compared them to 48K recordings?
     
  3. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    James-
    The clips were all originally 48k. The same audio files were put through the Dolby encoder first at 48k output, and then with the Advanced 96k Upsampling output enabled. The latest version of Dolby's Media Producer encoding software will allow encoders who have 96k files to do a 96k encode (no upconversion needed) with the option of using the same apodizing filter if they want to.
     
  4. Douglas Monce

    Douglas Monce Producer

    Joined:
    Nov 16, 2006
    Messages:
    5,514
    Likes Received:
    12
    Apodizing. Hmmmmm Sounds like the audio equivalent to motion smoothing. Great if you like your images, or in this case audio, processed and artificial sounding.
    I'll take the original, unmolested audio thanks. If they want in the future to offer REAL original 96k audio that started out that way, then they might have something. I'm sorry but anytime the equipment starts altering the signal to "improve" it, that's a BAD thing! Leave the audio as the sound designers and mixers intended. it is in no need of "improvement". Improvement, meaning it no longer sounds the way it did when they mixed it.
    Doug
     
    andySu likes this.
  5. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    Actually its nothing like that. Preringing is an unnatural artifact introduced as a byproduct of the D to A conversion. Its not present in an analog recording (assuming a complete analog path). Meridian (and now Dolby) use the apodizing filter to remove the distortion that was artificially introduced, to make it sound more like the original analog source. It sounds much more natural having been run through the filter.
     
  6. Douglas Monce

    Douglas Monce Producer

    Joined:
    Nov 16, 2006
    Messages:
    5,514
    Likes Received:
    12
    And considering that modern soundtracks are, with very rare exception, digital from start to finish, and there is no analog path, this is completely irrelevant.
    Doug
     
  7. Douglas Monce

    Douglas Monce Producer

    Joined:
    Nov 16, 2006
    Messages:
    5,514
    Likes Received:
    12
     
  8. Hal F

    Hal F Stunt Coordinator

    Joined:
    Feb 5, 2005
    Messages:
    127
    Likes Received:
    0
    I believe you will always need a D to A conversion in order to listen to it. Speakers are analog.
    In any case, unless this was a true double-blind test the results should be treated with a great deal of skepticism. And it doesn't sound like it was based on what Adam described.
     
  9. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    Yes I understand that there are probably no analog mixes today (at least one stage would be digital), my point is that every mix has artifacts that aren't supposed to be there that are by products of the "brick wall filter" that are used in every A to D conversion. I don't see how removing something that's not supposed to be there is "smoothing" out the sound. They are simply taking the distortion before the sound, and moving it behind the sound where it will be masked by your brain.
     
  10. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    I was making a reference to the "pre-digital world". Now everything starts as analog (dialog, sound effects, music etc.) is converted to digital and stays digital through the entire path to until the amplifier/speaker.

    I think skepticism is a strong word. It wasn't a blind test (except for the clips without video ), but I think the impressions are still valid. Unfortunately there are no plans for any type of 48/96 comparison clips to be made available to the public.
     
  11. moovtune

    moovtune Stunt Coordinator

    Joined:
    May 3, 2010
    Messages:
    150
    Likes Received:
    55
    Location:
    Los Angeles
    Real Name:
    James
    So wouldn't applying this filter on the original 48K files have the same effect - why is it necessary to upsample them to 96K? Did they take the original 48K digital files and encode them to 48K True HD with the same filter and let you hear those or is it strictly a 96K filter?
     
  12. Chuck Anstey

    Chuck Anstey Screenwriter

    Joined:
    Nov 10, 1998
    Messages:
    1,624
    Likes Received:
    102
    Real Name:
    Chuck Anstey
    The sound engineers created the problem in the first place when doing the initial A/D conversion. They introduced the ringing artifact that makes it no longer sound like the original analog signal when converted from digital back to analog for amplification. Dolby is trying to fix a known artifact of incorrectly filtered A/D conversion that manifests itself during the D/A conversion to reproduce the original sound. This wouldn't be an issue of if the original recordings were sampled at 96KHz or were properly filtered for 48KHz.
    After fixing the ringing artifact, which in this case is at its most basic is an under-sampling issue, to keep it from being recreated in the D/A conversion at your receiver the sampling rate must be increased or you still wind up with frequencies too high for the sampling rate and thus pre-ringing.
     
  13. Hal F

    Hal F Stunt Coordinator

    Joined:
    Feb 5, 2005
    Messages:
    127
    Likes Received:
    0
    I actually think 'skepticism' is a rather mild word to use in the circumstances. And until Dolby is willing to set up some double-blind tests conducted by independent experts one is certainly entitled to that skepticism.
    If you do some reading in the psychoacoustic literature you will soon learn how easy it is to be deceived by one's audio impressions. One book that I would highly recommend is: "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms" by Floyd Toole.
     
  14. Douglas Monce

    Douglas Monce Producer

    Joined:
    Nov 16, 2006
    Messages:
    5,514
    Likes Received:
    12
    Unless this filter is applied at the mixing stage so that the sound mixers can hear what it is doing to their audio, you are NOT hearing what the sound designers and mixers heard or intended. This is the audio equivalent of a video "noise reduction" filter on a TV, plain and simple.
    Doug
     
  15. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    I can see the value in doing that if you were going to need to buy new hardware to take advantage of it, but considering that some upcoming releases that you buy are going to have it, it will playback on your existing gear and won't cost you a dime it seems a bit overkill. I have been to a lot of trade shows and other industry events where various brands have been running demos comparing their cables, power conditioners, etc, etc against the competition with rigged tests. I believe that Dolby was above board with their tests, but the fact that it it wasn't a double blind test was why the thread was titled " first impressions".
     
  16. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    I would agree if there was a switch on the mixing board to "override AtoD conversion artifacts", but there never had been so mixers haven't had a choice, and until Meridian developed the filter a few years ago there was no way to remove this. It's one of the things you have to accept when working in the digital world. Dolby provided testimonials from several mixers that they were looking forward to using it on their mixes. Keep in mind that if a Director or Sound Designer doesn't like it or want it they don't have to push the button when it comes time to encode the audio and can stick with the artifact.
     
  17. Hal F

    Hal F Stunt Coordinator

    Joined:
    Feb 5, 2005
    Messages:
    127
    Likes Received:
    0
    That is assuming that there would be an audio difference with the speakers one currently has.
    I see little point in getting enthused about this until some more reliable data is available that can back up Dolby's claims..
     
  18. Brian Kidd

    Brian Kidd Screenwriter
    Supporter

    Joined:
    Nov 14, 2000
    Messages:
    1,913
    Likes Received:
    219
    I assume you mean THE DARK KNIGHT? :)
    Thanks for the report. It seems to be an interesting development, but I doubt it will cause many to go out and purchase new equipment. It will be nice if hardware manufacturers start including it and don't charge a substantial premium for it.
     
  19. Adam Gregorich

    Owner

    Joined:
    Nov 20, 1999
    Messages:
    15,672
    Likes Received:
    671
    Location:
    The Other Washington
    Real Name:
    Adam
    I'll have to double check my Batman titles later today. That's the beauty of it. If you have a BD player or receiver that can decode Dolby TrueHD today, you don't need to purchase anything. Just about every device that can decode Dolby TrueHD will have DACs capable of handling 96K. It is all done on the encoding side so there is nothing for the HW manufactures to include.
     
  20. moovtune

    moovtune Stunt Coordinator

    Joined:
    May 3, 2010
    Messages:
    150
    Likes Received:
    55
    Location:
    Los Angeles
    Real Name:
    James
    I'm not following how the mixers will be able to audition the results of this process. The encoding to True HD is not done on the mixing stage but I assume by the editors in the home video department in prep for a Blu-Ray release. And if it's a filter that works best with higher sample rates (96K), then the mixers, working at 48K, wouldn't have access to the improvement during their work on the mix. That would only come later during the encode process right? Well after the mix is done.
     

Share This Page