iZotope RX10

[noiseboyuk]

Perhaps it's significant that there is no thread on RX10 a week after release. How the mighty fall... when Pro Tools Expert gives it the thumbs down you know you're in trouble.

I downloaded the demo. Speech to Text is useless. But the new Spectral Recovery did occasionally produce great results in the low end on speech. It's hit and miss (as is so often the case in NR), it seemed to me that when fixing a gradual LF roll-off it coped better than an abrupt one (though I also had to massage lower mids with EQ afterwards. Listen out for general graunch along with boosting the fundamentals, sometimes its minimal, sometimes overwhelming.

But overall that's two majorly disappointing RX releases in a row. Meanwhile Acon go from strength to strength, and Waves Clarity was a remarkable surprise from out of nowhere. What else are people turning too these days?

[smurfyou]

Accentize DeRoom Pro, Chameleon, and VoiceGate.

Spectral Layers pro.

Cedar of course.

[Eric Lambert]

Izotope has been dropping the ball quite a lot in recent time. I'll be watching this thread for alternates.

[Reckless Erik]

I think their focus is now consumer/podcast. Even though I use RX every day I find that little of what they have released in the last year is a game changer for post.

[musicman691]

Watched a couple of YouTube vids on RX10 and the general consensus was to hold off on it. Speech recognition only worked on American English and some American accents but not all. Also it requires really clean audio. And the rest of the so-called updates weren't thought of all that highly either. For those on Macs you need a minimum of OSX 10.15.6.


Ozone 10 has also just dropped. Some cute stuff but nothing really in the meat and potatoes realm.

[audiolex1]

Yeah, seems pretty not feature heavy. I know it says 10.15.7, but it is like PT where it just isn't being supported but should work fine?

[noiseboyuk]

Originally Posted by Reckless Erik

I think their focus is now consumer/podcast. Even though I use RX every day I find that little of what they have released in the last year is a game changer for post.

I think there was a big shift between 8 and 9. Until 9, each release was a hefty update with major new features, many of which were trailblazing. It also had a hefty update price to match of $299 (sales would come eventually, of course).

With 9, new features were notably thin on the ground, and the update price dropped to $149. Now in an of itself that’s not a terrible thing, but I don’t think 9 and 10 together really add up to one old-style update. The features are either irrelevant or poorly implemented or both, all at a time when competitors are delivering. I don’t know if it is to do with the Soundwide thing or what, but either a lack of resources or poor management are really biting.

I’m really quite sore that I spent a lot of time with iZotope support after the release of RX9 to demonstrate that Dialog Isolate’s new dialog detection algorithm had major issues, performing worse than RX8. They thanked me profusely and agreed they should look into it. They’ve done literally nothing. It speaks volumes to me - pouring resources into a clumsy gimmick like audio-to-text and neglecting the core business.

As I said in the OP, I’ve tried the new LF mode in Spectral Repair and when it works its really good. I’ve thought for a long time that it should be possible to detect and isolate the speech fundamentals and harmonics, and boost (even generate) those that are missing. But I feel they just stopped this mid-development. Quite subtle changes in the source can throw it off completely, and it has a tendency to just boost random bass, resulting in horrible sounding growling (that Dialog Isolate can’t get rid of, incidentally).

[LDS]

Originally Posted by noiseboyuk

I think there was a big shift between 8 and 9.

It all coincides with the announcement of their partnership with Native Instruments after a big technology investment firm started pouring the dollars in. Plugin Alliance & Brainworx have since joined, and much of their voucher system was reformed in the months that followed.

Check out their website.

"Soundwide on track to becoming the leading audio and music creation ecosystem".

Diverse and Integrated.

Synergy.

Empower Communities.

Collective Vision.

Serving the creator community.


Vomit. It reads like someone accidentally copied a buzzfeed '40 most meaningless business words that need to die' article. This is the future, hyped to death, endlessly cross promoted, artist endorsed, with an integrated TikTok & Instagram share button right there in the menu bar. You'll be famous overnight, doncha know?


Maybe I am just getting cynical...

[musicman691]

Originally Posted by noiseboyuk

I’m really quite sore that I spent a lot of time with iZotope support after the release of RX9 to demonstrate that Dialog Isolate’s new dialog detection algorithm had major issues, performing worse than RX8. They thanked me profusely and agreed they should look into it. They’ve done literally nothing. It speaks volumes to me - pouring resources into a clumsy gimmick like audio-to-text and neglecting the core business.

I wonder if what you are seeing is an inability to deal with other than American accents which is something that YT vids mention in relation to RX10. Old code that hasn't been fixed. It's hilarious (not) watching a vid where the reviewer is trying the new speech detection and the words it comes up with are really bad. Sort of like when closed captioning first came on the scene. Yet YouTube's closed captioning seems to work a treat across languages and accents.

[gives]

Originally Posted by Eric Lambert

Izotope has been dropping the ball quite a lot in recent time. I'll be watching this thread for alternates.

Copy that. They need an Algorithm update.