The gold standard for audio repair has improved yet again

Share This Post

For example, Remediation Assistant, which already supported machine learning in the last three versions, now gets more intelligence and suggests fixes in more areas. I think his one new feature in the RX 10 Advanced will benefit everyone working in the broadcast and film industry. It’s a feature that enables text navigation in dialog editing.

iZotope’s RX application is the gold standard in audio restoration and enhancement. Erik Vlietinck tried the latest version. iZotope RX is one of the must-have apps for audio engineers. Luckily, the new version of iZotope released a few days ago includes a set of features that allow you to further edit your dialogue and monologues Efficient.

Highlights

  • If you’ve been around text-to-speech apps lately, you’ll notice that the feature’s current status isn’t bug-free: the RX 10 isn’t 99.99% accurate, and it doesn’t need to be. Recognized text is just for hacking large files with sounds like words. For example, in iZotope’s test file, a man and a woman are in a studio talking about furniture in American English. By my calculations, the transcription engine misunderstood about 10% of the dialogue.

  • Text navigation transcribes the spoken text of up to eight speakers and separates them using a color-coded “trail” for each recognized speaker, just above the spectrogram and in the sidebar containing the search field. to be able to distinguish between You can search for words using the search field. For example, you can quickly move to parts of the dialogue that need de-essing the most, leaving the other voice input alone. This is the coolest new feature in RX 10 Advanced and in future versions we may be able to export the transcription to a text file. 

In both tests there was a lot of background noise beyond the speakers. It ranged from sounds of crashes, devices turning on and off, and objects being thrown on the desk. This background noise led to a third speaker track. It’s a mystery (or bug) why the first test revealed a second speaker, but there’s no evidence.

For my own test file, in my best English with a Belgian accent, I transcribed a section of Wikipedia about Merlin, the magician in Arthurian legend. The transcription machine misread about 15% (again, my guess). RX 10 speaker identification is also not 100% accurate yet. While I was alone in the room, RX claimed there were two of us, one with a blue dot next to his ID and one with a yellow dot next to his ID, but No second trace.In another file with 2 people in the same room, I thought RX was 3 people.

Related Posts

As Stadia greets, Luna welcomes new games for October

While Luna is still around and Amazon recently unveiled...

Sega presented Sangokushi Taisen, a blockchain game

Over the past few months, several major Japanese publishers,...

Management change announced by NaaS Technology Inc

Alex Wu served as Interim Chief Financial Officer of...

health and communications leaders join the Oregon board Tech Foundation

Two new members have just joined the Oregon Tech...