Jump to content

Audionomix? cleans up production dialog...


Ty Ford

Recommended Posts

No.  I notice that they only gave us one very short sample of treated speech (that sounded very "sat-on" to me) before they fired up the reverb to camo what the track really sounds like.  It seems like it costing $1k and that it's a stand-alone only (no auto-xfer between a DAW and the app) there are other ways to do this work that are more compato with current workflows.

Link to comment
Share on other sites

  • 2 weeks later...

Never used this one but Melodyne does a similar thing and the result is generally tinny and missing frequencies.   It has to be hidden in a mix to be usable even for music.  RX is still probably more useful in general, plus I don't really trust how audionamix does advertising/promotion.

Link to comment
Share on other sites

  • 2 weeks later...

Just a quick update*.

I reached out to them as a user/journalist/educator back in early March, when they announced what was described as nothing short of miraculous. I wanted to know whether they could provide a more revealing demo, or process a track of mine, or give me a one-week NFR license to evaluate. My concerns were that their music app appeared to be pitch/harmonic based, and that's nearly impossible with speech (pitch changes too quickly for most tracking and the best portable vocoders there days -- aka cellphones -- have quantization; on top of that, the harmonic structure depends on the vowel and can be constantly gliding). So  also wanted whatever details they could release, and would sign an NDA for that part if needed.

 I got a call back from a woman who identified herself as doing their marketing. She promised I could have a talk with one of their developers... when he was available... probably after NAB.

I saved the notes/contact info, and tried again a few weeks after NAB. No response this time.

I poked around on their site, and it appears as though this really isn't a native plugin. Instead it requires net access, phones home with the track, and just leaves a proxy on your machine while it's working. No idea how much computer it requires at their lab, or how much human tweaking is necessary. Haven't been able to verify this (or get it denied) with anyone from Audionomix. 

Since then, nothing.

My take: something like this will occur. We're getting close enough with natural speech recognition that includes DSP noise reduction, and some form of phonemic analysis (Google Home / Alexa anyone?). We're not quite there with natural speech synthesis mimicking real-time voices, but it'll come. When those two converge: it you can't clean the production dialog, just re-create it without the noise. True ADR!

Just, probably, not on this year's Christmas List.

--

*I was offline for a while. Getting married, then having ortho surgery. Both completely successful.

 

 

 

Link to comment
Share on other sites

15 minutes ago, Jay Rose said:

Just a quick update*.

I reached out to them as a user/journalist/educator back in early March, when they announced what was described as nothing short of miraculous. I wanted to know whether they could provide a more revealing demo, or process a track of mine, or give me a one-week NFR license to evaluate. My concerns were that their music app appeared to be pitch/harmonic based, and that's nearly impossible with speech (pitch changes too quickly for most tracking and the best portable vocoders there days -- aka cellphones -- have quantization; on top of that, the harmonic structure depends on the vowel and can be constantly gliding). So  also wanted whatever details they could release, and would sign an NDA for that part if needed.

 I got a call back from a woman who identified herself as doing their marketing. She promised I could have a talk with one of their developers... when he was available... probably after NAB.

I saved the notes/contact info, and tried again a few weeks after NAB. No response this time.

I poked around on their site, and it appears as though this really isn't a native plugin. Instead it requires net access, phones home with the track, and just leaves a proxy on your machine while it's working. No idea how much computer it requires at their lab, or how much human tweaking is necessary. Haven't been able to verify this (or get it denied) with anyone from Audionomix. 

Since then, nothing.

My take: something like this will occur. We're getting close enough with natural speech recognition that includes DSP noise reduction, and some form of phonemic analysis (Google Home / Alexa anyone?). We're not quite there with natural speech synthesis mimicking real-time voices, but it'll come. When those two converge: it you can't clean the production dialog, just re-create it without the noise. True ADR!

Just, probably, not on this year's Christmas List.

--

*I was offline for a while. Getting married, then having ortho surgery. Both completely successful.

 

 

 

Thank you for a very interesting post and congratulations on getting married.

d rosen

Link to comment
Share on other sites

For this to work in live-fire post the round-trip will have to be wicked fast--at least as fast as doing the work on a local computer.  I think the increases in speed of computers and apps you can buy might very well make this irrelevant.  Are they running a big mainframe or server farm?  This idea of sending the sound off to another place to be "fixed" runs counter to what I see as current trends and philosophy in computing, which is all about devolving power to the individual local device.

Link to comment
Share on other sites

There was a piece on BBC Radio 4 a few days ago about the latest state of speech synthesis. They had a computer that was doing a pretty passable job of synthesising Trump. Ift was not totally convincing, but if you added a bit of BG noise and made it as if through a phone it was pretty much indistinguishable from the real thing. One of the points that they made was that it is now close enough, that from now on no one should necessarily believe that anyone in particular has actually said anything, unless you are in the room and hear/see them say it ;-)

Happy days.....

Simon B

7 minutes ago, Philip Perkins said:

 I think the increases in speed of computers and apps you can buy might very well make this irrelevant.  Are they running a big mainframe or server farm?  This idea of sending the sound off to another place to be "fixed" runs counter to what I see as current trends and philosophy in computing, which is all about devolving power to the individual local device.

Can you imagine the big studios being cool with the dialogue editor sending off lines from their not yet finished film for some online service to process? sb

Link to comment
Share on other sites

no one should necessarily believe that anyone in particular has actually said anything, unless you are in the room and hear/see them say it

...and are close enough to hear them say it acoustically, and see their lips move in person. Between what we're getting to in synthesized speech, and the lip-sync apps our CGI friends develop, you won't be able to trust a giant on-screen Trump at a giant rally even if you're in the live audience.

 

 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...