Restoration Suite 2 — a Giant Leap in Audio Quality

[stian]

In the ancouncement as of Oct 16, 2019 you mentioned that the Restauration Suite 2 will be includes in the "upcoming Acoustica Premium Edition 7.2.
Two days later Stian wrote "Yes, we're working hard on v7.2. It's been slightly delayed, but we expect to release in a month or so. We'll need to get it out quickly for full macOS Catalina support as mentioned earlier."

Now 4 month later there is no version 7.2; when will it come???

Peter

I'm sorry for the delay. Getting all our plug-ins ready for macOS Catalina took longer than expected. The release of the "spleeter" source separation libraries also came as a surprise and I thought it would be worth waiting a little extra for having that functionality inside Acoustica. Spleeter lets you split a complete mix into stems (you can read more here: https://deezer.io/releasing-spleeter-de ... 88985e797e). The technology is still artifact prone, but spleeter is better or up-to-par with all the commercial offerings to my ears. The open source implementation is very hard to use (you need to install python and use it from the command line), doesn't offer real-time preview, only supports stereo at 44.1 kHz and the output is band-limited at 11 kHz. This will be very stream-lined in Acoustica without the mentioned limitations of the open source version. I've also combined Spleeter with some of the techniques from our denoiser in a new processor called "Dialogue Extract". This will be demonstrated at NAMM next week and the current ETA for the official release is end of February.

Best,
Stian

[williram]

Are you using the plug-ins inside Acoustica or in another host? Normally, the plug-in reports the delay the host compensates for it by delaying all other signal paths.

Firstly, thanks for the quick reply and also awesome software!

I am using the plug-in's in FL Studio 20 and am thoroughly confused about the latency because it doesn't happen all the time. However, the issue does seem to be that of the host, FL Studio.
Since a few other problems are occurring. Like that of a different resulted sound and latency effect when clicking on a different mixer track that the acon digital plug-ins are being used on.

The latency problem occurs when using the "denoise2" and "dehum2" plug-ins inside of Acoustica 7 Premium Edition as well when inside a multi-track session (which has never been able to export on to any file without extreme and very frequent artifacts in the audio.

[nathanj]

I’ve also combined Spleeter with some of the techniques from our denoiser in a new processor called "Dialogue Extract".

This sounds amazing! Thank you for updating your already awesome software with such significant improvements. I love this suite.

[Gevatter]

I have the spleeter funktion already in iZotope RX7 and I am not really convinced about the results!

Regard Peter

[XComposer]

I read that the open source implementation (which I don't have) can be "trained" to learn to separate new kinds of signals and new different instruments. It would be nice to have this feature implemented also in Acoustica, but I imagine that it's complicated.

[stian]

I have the spleeter funktion already in iZotope RX7 and I am not really convinced about the results!

RX7 doesn't use spleeter, but yes -- the concept is the same.

Best,
Stian

[stian]

I read that the open source implementation (which I don't have) can be "trained" to learn to separate new kinds of signals and new different instruments. It would be nice to have this feature implemented also in Acoustica, but I imagine that it's complicated.

It's a good idea, but unfortunately, training these networks is a tremendous work and you would need a very large library of recordings split into separate stems. Deezer, the company that developed spleeter, has access to a large library as it's a music streaming service, but with the current technology, it won't be possible for end users to do the model training, I'm afraid.

Best,
Stian

[stian]

I am using the plug-in's in FL Studio 20 and am thoroughly confused about the latency because it doesn't happen all the time. However, the issue does seem to be that of the host, FL Studio.
Since a few other problems are occurring. Like that of a different resulted sound and latency effect when clicking on a different mixer track that the acon digital plug-ins are being used on.

To be honest, I'm not sure why it doesn't work in FL Studio 20. Maybe the latency is longer that FL Studio supports.

The latency problem occurs when using the "denoise2" and "dehum2" plug-ins inside of Acoustica 7 Premium Edition as well when inside a multi-track session (which has never been able to export on to any file without extreme and very frequent artifacts in the audio.

I'm sorry to hear that. I looked into the issue, and unfortunately, there's a glitch in the latency compensation in the multitrack editor in Acoustica. The problem depends on the latency, so only some plug-ins are affected. The internal tools are properly compensated, so problem is restricted to external plug-ins. I've fixed this already in the code base and it will corrected in v7.2.

Best,
Stian

[Gevatter]

Now I can see, it is end of February.
Unfortuanatly, I can't see Acoustica 7.2

[stian]

Now I can see, it is end of February.
Unfortuanatly, I can't see Acoustica 7.2

Thanks for the interest and I apologize for the delay. Release is very near now, however, and we're in the final beta testing stage.

Best,
Stian

[stian]

Regarding the Acoustica 7.2 release date, please stay tuned on March 11th.

Best,
Stian

[ToxMox]

Is there a way to use the Extract Dialog feature as a plugin in my DAW or is that feature planned? This is what is stopping me from buying/crossgrading.

Thanks!

[stian]

Is there a way to use the Extract Dialog feature as a plugin in my DAW or is that feature planned? This is what is stopping me from buying/crossgrading.

Thanks!

Thanks for the interest! It's not possible to create a plug-in version with the current AI models as they require audio chunks of around 10 seconds. Acoustica pre-renders in a background thread to give the feeling of realtime processing. However, this is an area that we are focusing on and we hope to be able to provide a real-time solution with an acceptable latency. This won't happen in the very near future, though.

Best,
Stian

[ToxMox]

Thanks for the interest! It's not possible to create a plug-in version with the current AI models as they require audio chunks of around 10 seconds. Acoustica pre-renders in a background thread to give the feeling of realtime processing. However, this is an area that we are focusing on and we hope to be able to provide a real-time solution with an acceptable latency. This won't happen in the very near future, though.

Best,
Stian

Ah ok, understood. Thank you.