Jump to content

Jay Rose

Members
  • Posts

    1,295
  • Joined

  • Last visited

  • Days Won

    1

Everything posted by Jay Rose

  1. Ah... those precious minutes between 'roll' and 'action' are where I often gather samples for my roomtone generator.
  2. And furthermore... I sent the TV Tech article to a local SMPTE friend of mine. His reply which got me thinking about what's really happening: There are two "problems", both economic: 1) Samsung (and others) want to be able to brag about 8k, so they can sell another generation of sets. We've reached the practical limit of how big a screen folks can put in their living rooms, and today's sets don't wear out. So manufacturers need some new feature to drive next year's sales. Ditto Netflix, Amazon, AppleTV, Disney, and all the other new streamers: they need some competitive hook to say they're better than the others. How about 8k? 2) Bandwidth isn't free. And it'll get even more valuable, now that the FCC has decided that Net Neutrality is a commie plot. So those streamers will be looking for some way to squeeze their vaunted 8k down a 4kbps pipe. If a cheap single-chip NN in the viewers' sets can turn SD into 8k, it'll be a win-win! At least, for everybody other than filmmakers and viewers.
  3. This is about video, but the results will surely filter down to our department(s): Engineers are teaching upscalers to recognize different kinds of scenes, and fill in missing pixels by using neural networks. Samsung is trying to build this AI into consumer sets. TV Tech article. Ultimately of course this depends on the quality of the AI training. But low-cost engines are getting faster, and there's a lot of training capacity available. So it's sure to grow. Here's the thing: if this becomes the standard in a few years, and content providers start using more compression to save money, because they know the viewers have AI upscaling... it'll be almost impossible for a director to come up with a new look for a show! Once a show goes into distribution, if consumer sets haven't been trained for that look, the sets will force the new footage to match something they already know! Think of it as 'mandatory auto-tune' wrecking all the great jazz singers of the past...
  4. Where to start? If they've already got a DSP system with good echo cancellation, why not just take a feed from it? Or if you have to mic and mix your own: Earbuds (or IFB) on the local talent, feeding a mix minus of just the dx audio. Or a tiny cue speaker for the dx, low level in the main conference room, and automatic mixer to duck local when it goes below a threshold. Clarity -- and keeping the dx talent from going nuts with delay -- is more important than 'client comfort' at the shoot. And, of course... while a large part of this kind of echo can be coding delay, it always helps to do something about the room acoustics.
  5. I don't know of any 24-bit field recorder that actually achieves 144 dB dynamic range. (Zaxcom, for example, claims 120 dB.) The limitation is the analog and ADC circuits. More bits in the file will just be random junk or zeros. You're also limited by the dynamic range of the mic. And of the set... If an actor is projecting at 85 dBC, it's only that far from the theoretical threshold of hearing. OTOH, more bits are definitely an advantage in post. Padding the depth with actual zeros can give you a lot more room to avoid rounding errors when processing. Then dither down when you've got a final mix.
  6. I imagine you'd need The Hulk to run boom. It's easy to come up with materials that absorb sound, in the sense of letting it pass through the material's surface and not be reflected. The problem is, what do you do with that energy after you've absorbed it? Something's going to have to vibrate, letting the energy turn into heat, followed by a barrier so the vibrations don't emerge on the other side. That's usually done either by mass or decoupling, either of which will make your zeppelin a lot bigger and heavier. (You can also use a vacuum, but I'm not sure how you'd accomplish it.) If you don't want sound to get to the mic, you have to reflect it away rather than absorb it. Either of which solution brings us back to the OP's idea of stopping the interference tube from doing its job, giving us an omni.
  7. If you want an upper-mid boost, suggest you get a flat recording and do the tweaking in post. Where you've got much better monitoring that headphones on the set, and you've got an undo button. If you record with that boost and a low-cut for brightness and grit, and then you want to put the lows back in post... they won't be there. No amount of EQ will help. BTW, I wrote that book Jim is recommending. He doesn't get a commission. In fact, all I get is about $4 per copy, after Amazon and the publisher take their share. But it is used by a lot of US film programs, and has been favorably reviewed by some A-listers. I wrote it to share what I've learned, both from experience and from friends in the industry. Details and samples - so you can judge for yourself - at greatsound.info.
  8. FWIW, Carl Countryman once told me that the only reason for a cardioid lav is if a live-performance talent is in front of a stack. Otherwise, you're dealing with all sorts of problems (direction, plosives, etc) that are much nicer in an omni. I've also had excellent luck with active reality in very noisy exteriors with the near-invisible mini headset mics. Like the Countryman E6, but there are others. Production has to sign off on the look -- which can be barely noticeable, except in CU -- but do they want good sound?...
  9. Don't believe a mic has a frame of view, the way a lens does. Things to the sides and rear will get picked up, even with a $$$ directional mic. Don't ever believe the ads about how directional a mic is, or whether it can 'zoom'. The only meaningful spec of this type is a set of polar patterns... which have been measured in a proper acoustic space. In your room, even those patterns are likely to be off. In your situation, the best solution is a tiny head mic, placed a fraction of an inch from the side of your mouth. That lets physics work for you, rather than against you, by changing the relative distances to your voice and the speakers. And you might consider running your speakers a little softer... (BTW, if a dynamic gives you much more noise than an equivalent condenser, you need a better preamp. In terms of pure physics, dynamics generate a lot less noise than condensers. But with modern equipment, both of their noises are so low not to worry about.)
  10. I'm prepping a 2-hour docy that has an important one-shot premiere in a historic theater... in ten days. (Needless to say, it's not locked yet.) Of the two hours, about five minutes has stereo music or sfx. The rest is archive footage and new interviews. The theater can handle DCP, but the producer can't take the time for a lab. He's leaving my studio with a hard drive under his arm, hopping on a plane with the editor and his laptop, and they'll be syncing and compressing as necessary during the flight to the venue. The theater can also take BluRay with AC3, or play a stereo or LCR track with picture from a hard drive. I want to leave as little possible room for error at the venue. But I'm also concerned that in a 'historic' theater, having them use just their screen center channel for 96% of the movie won't fill the room. Or it will cause problems for people at the extreme sides of the front rows. Ideas? Mix as LCR, but also send my dialog to L and R down 6 dB or so (to keep things centered) at the same time? Mix stereo with a phantom center and hope for the best? After this premiere, they'll be shortening for theatrical /streaming and we'll have a chance to give the track more thought. Fortunately, picture is someone else's worry.
  11. Congratulations to her! I can tell from your posts that writing skills run in your family. Well, apparently they run at Olympic level for your daughter!
  12. I agree that the historic bias against women's voices, particularly when it comes to 'authority' as a characteristic, is a horrible thing. But I suspect the reasons are cultural rather than technological. ...or at least, it leaves me to question some of the stuff that's quoted in the article. - The fundamental range of male voices is lower, but that would hurt men more than women when early technology considered 100 Hz to be a decent low-end cutoff. That fundamental range -- a buzz formed by air past the vocal folds -- is where most of speech power lives for both men and women, in terms of SPL and duration. But things aren't very intelligible if that's the only band we're listening to. - The range above 1k5 or so and where intelligibility lives, and about the same for adult men and women. It's determined by the size of the head, which affects the head resonators (mostly little wetware Helmholtzes, formed by the tongue moving against cavities in the mouth) that reinforce harmonics of the fundamental buzz. These harmonics are why you can sing different vowels on the same pitch, in case you've ever wondered. That >1k5 band is also where articulators form the critical differences between different consonant pairs*. Prove it to yourself: you can varispeed a male voice up to make a kid (or small cartoon animal), raising all the frequencies by the same factor and implicitly making the 'head smaller'. But you can't pitch shift a man into a woman. Unless you use a 'formant corrected' pitch shifter, which uses a crossover to keep those higher frequencies from being raised by the shift on the lower ones. Or see it for yourself: I've put together a web page with spectrograms and audio samples comparing male and female announcer voices and a few different musical styles, both wide band and separated into smaller bands. You can see and hear what's happening at different frequencies, and how it all relates to power and intelligibility. You may be surprised at how similar male and female voices are, when they're delivering in the same style. And some related music-history trivia, from back when I was getting my degree: in classic operas, heroes and military leaders were tenor rather than basso. That's because traditionally, higher voices were know for carrying better on a battlefield. -- * "Consonant pairs"? Most consonants come in voice and unvoiced pairs, formed by the same mouth movement in an airflow but with or without the vocal fold buzz. /p/ and /b/ are that kind of pair. So are /s/ and /z/, or /k/ and /g/. (That degree I studied for was in speech science. It's given me a nifty a career as an editor and processing designer.)
  13. Jay Rose

    Best Lavs?

    Don't confuse lav with wireless. Unless you start spending $$$ just on transmitter and receiver, a wire will always sound better. And unless you've got that kind of a rig, it's worth making the effort to run a wire on scenes that can't be boomed. Boom v Lav is a totally different question. It has to do with both where you're shooting and what the film is trying to do. As a postie, I can say with absolute certainty: a well placed boom AND a good sounding lav is the best way to record a track. Even in narrative films where boom will be the hero, having a lav can solve a lot of problems.
  14. BTW, on those thin shots... there was no reason to use the mixer's crappy radios. It was a noisy environment, so lavs were a necessary backup. But both characters were seated for that entire scene, never moving, and shot from waist up: they could have been wired. (No idea who the mixer was, and never bothered checking IMDB after the fact. I suspect he might not have known lavs even can be wired!)
  15. You will never stop producers from shopping for the lowest price, even if it hurts their film. I commented to the producer on one project that production tracks from the second unit stuff sounded incredibly thin... to the point that there was nothing to equalize up. His responses? a) I got this guy for $350 a day with full kit! and b) You're supposed to fix it.
  16. If you’ve got the capital, it’s what I call a “spreadsheet decision”. Lease? What’s the total of your payments over time. Plus any cost to buy outright at the end of the term. Less how much the money you didn’t spend upfront would have earned you (here’s where the recursive feature of a spreadsheet really comes in handy, since that principal decreases with each lease payment). Buy? What’s the best cash deal you can make. Less any resale value if you don’t intend to keep it. If if you don’t have the capital, you still might be able to get a better deal ( or used or demo gear, often not available on lease) by borrowing cash, possibly secured by other gear. Often the interest rate will be less... and the loan may have better early termination terms than a lease, if you get some extra bucks and want to buy it out. In my experience, the tax implications are often minor. You can deduct lease payments... but you can depreciate owned equipment. Under current US law, you may be able to expense the entire price of owned equipment the first year. But I’m not a lawyer or tax accountant... the rules for you - even in the US - may be different.
  17. So now I have to doubt everything I've learned about Baseball and the Civil War? 😉 -------------------------------------------------------------- Update: About six hours after I sent correx to the Times' news desk, the article quietly changed to:
  18. Today's New York Times has a nice article about early country music recording and its stars. Also this: ...which would surely amaze the engineers trying to make magnetic recording a practical music medium before we captured the German innovation of high-frequency bias after WWII.
  19. Thanks, Philip. I’ll check with the production how it’s going to be screened. If a computer file, yes, LR makes sense. But if he’s planning to come in with a DVD, I’m leaning toward Ac3 LCR. I’ve seen too many LR DVDs badly de-matrixed, including one of this director’s (with his $ partner hard left in the front row, complaining about the mix).
  20. Interesting thought, LD. Narrative or editorial holes aren't my department, fortunately. The director and editor are very good and we've all worked together before. Playing with historical stereo fields might be an idea. I'd usually do that with bandwidth. It'll be a tough sell to add crowds, however: director has already told me he doesn't want any sweetening on the historic media, and scoring will be very sparse. (Credibility is very important to him on this one.) I did something similar on a music docy for the same guy about seven years ago. This traced the history of a very important club in Harvard Square, and included half-century old mono club recordings of Joan Baez, Bob Dylan, etc along with contemporary interviews and performances. Often I'd do something like letting the narrator introduce an archive performance, then dissolve to the mono of Joan center plus LR. After a verse or so, I'd fade Joan into a heavily stereo-simulated version to clear the center while the narrator came back. The stereo sim was mostly complementary combs rather than verb: I didn't want to distance her, just make her more of a 'surrounding' memory. (Film's IMDB. The whole thing is on Amazon Prime; there's a heavily compressed clip of one of the Joan interviews on YouTube. Hearing it again, I sure wish I'd had RX7 in those days...)
  21. I haven't heard the clips yet, but from what I know of the modern interviews, they're all over the map: different offices, different crews, different acoustics. Probably a bit of (mono) verb already on them. I can probably de-reverb the historic stuff in the interests of cleaning it, but my biggest goal with the modern stuff has to be consistency. But you raise an interesting point about LCR being identical. I might play with very small delays from C>R and C>L, slightly different for each, so that Haas forces the perceived source to the center of the screen. Everybody hears dialog, and the audience on extreme left and right of the screen hear less of the pre-signal coming from C so there's less of a Haas drawing them across. I'll have to play with a couple of DVD decoders to see if it messes them up further. (Yes: I know there's plenty of good reasons to keep dialog in the center only. And I know DCP is the right way to do this. But I'm trying to compensate for other realities here.)
  22. (Note on subject line: I'm referring to this current posting. General comments about my sanity may also be accurate, but I already know I'm cuckoo.) A producer that I like came to me with a semi-freebie: a two hour theatrical documentary about a 50-year-old piece of American history, with lots of contemporary interviews with the folks involved plus historic clips. He's also licensed some scoring from a 1960s mainstream feature as a contribution. I like the guy; he's given me real projects in the past, with real budgets and real schedules, and I want to do this one. Along with very little money, he has very little time. I'll get about eight days for dialog edit, premix, M&E, and remix. Then it gets one very visible screening. After that, there'll probably be time and bucks to take it apart and tweak. (Even if not... as I said, I want to do this one.) Here's the rub: 1) He's reluctant to do a DCP right now, mostly because of the lab time involved. He's expecting to grab my final mix, hop on a plane, and be doing compression while he's flying to the venue. I know from bitter experience that anything shy of a DCP can be mangled by a theater's DVD or similar playback. 2) He doesn't want sweetening in the historic clips; just original footage as best I can clean it. He doesn't want scoring under the interviews; just the folks' voices. There's very little narration. It appears the only music, other than main title and credit, will be during interstitials and chapter break titles. In other words: very little of this show is stereo. None of it is surround. Since this scoring is all archive, probably nothing will hit LFE. What I'm thinking about is mixing the dialog as 3-track mono. Same material on all 3 tracks, maybe -3dB on the center one. The only time L&R will be different is probably during the title and credit. I'm figuring this will give me the best chance of everybody in the front row of the theater hearing a decent track. If I just mix in stereo with phantom center, I'm worried some decoder will decide to suppress the dialog (it's happened before)... or the theater's L&R mains will never be on because everything will be matrixed to the center (ditto). If I encode for the guy's DVD from this LCR master, maybe there's a chance all three front speakers will be talking. And, yes, I'll still try to convince him to pull a DCP. If there's not time for it at the premiere, at least for future exhibitions. Thoughts?
  23. [quoted in its entirety] "Wrong" as a reaction to a specific statement? Are you plugging a song by Depeche Mode or Luh Kel? There are a couple of folks named Wrong in IMDB; are you trying to ping one of them? Or is this a comment about life in general?
  24. Kishor, I applaud your courage trying something new, and wish you luck. I dropped location recording years ago, in favor of post, and it was a good move for me. The thing is, this is a new venture for you. Before you go buying new gear, you should probably get an orientation about what modern post entails. Is there someone who can mentor you or let you apprentice while you're learning? There are a couple of good books on the subject, including mine. Having a ProTools rig that supports picture and lots of tracks, and knowing how to edit dialog on it is a good start. The alternative to a full PT rig seems to be Nuendo, and many posties (myself included) prefer it. There are also cheaper alternatives if you can take the time to work within their limitations. Buying a suite of plug-ins, however, might be a mistake. It depends on your business plan. If you're going after theatrical films or high-end TV, the most important tool -- along with your DAW -- is a room with accurate and predictable monitoring. If you don't have that, you'll more likely be just editing and not processing; quality decisions (even NR) are best left when you're got all the elements in place and are mixing in a good room. If you're aiming lower and will be mixing what you edit, you still need good monitors. You also need a fuller understanding of the entire process, which takes you back to learning: far better to buy what you know you need based on the projects you're getting, rather than guess and buy something because some other postie uses it in their workflow. With that warning: I use Nuendo and the full suite of post-oriented plugins that comes with it. My favorite third-party stuff is the iZotope RX Advanced suite, some Waves' psychoacoustic stuff (mostly harmonic effects), and the ones I cobble together in platforms like the ancient 32-bit SonicBirth. Since my work includes SFX design and placement and have a big library, I've been a SoundMiner enthusiast for more than a decade... but there are alternatives now. As a Mac user, I keep TwistedWave as my quickie 2-track editor/proofer, and SoundGrinder as universal conversion tool. These are just personal opinions, off the top of my head and honed by the specific jobs I'm doing. Expect to hear other opinions. Your ultimate rig may be very different.
  25. In either a plant or a human, that would indicate a persistent vegetative state.
×
×
  • Create New...