www.hairersoft.com

I'm afraid that support for third-party plugins can only be obtained from their respective vendors, please contact iZotope's customer support.

OK, I found the problem I think. There seems to be a bug in this function when "smart editing" is enabled in the "Edit" menu. Try disabling this and it should work again as expected.

Sorry, I missed that post.

I don't think that this function has changed in a pretty long time. Selection the portion you want to loop and then entering the length of the crossfade seems to work as expected for me in version 2.8.4.

Regarding Rosetta 2: yes absolutely, the latest version (2.8.4) of Amadeus Pro and Lite works perfectly (and fast). A native version would not be all that useful at the moment since most Audio Units have not been ported yet, so these would break and many users would have to run it under Rosetta anyw...

Yes, if you could send me a Dropbox link to martin@hairersoft.com (to keep the file private) together with instructions on how to reproduce the glitch that would be great.

This can happen if there was a glitch there that has been removed, since it happens that the way in which the waveform summary (which is what you see at large zoom levels) keeps track of the changes doesn't always faithfully reflect the changes of the waveform itself. Having such a large discrepancy...

Metal seems rather unstable under High Sierra; this is fixed in Catalina and Big Sur. (And wasn't a problem in Sierra either.)

By spinning the writing itself off to a separate thread I can get the write time down to about 3.8 seconds, same as Audacity. I guess that's getting pretty close to optimal. (Speeding up by a factor two isn't realistic since there's still a data pipeline that needs to be synced...) I'll probably inc...

I've profiled the writing. There is a smallish (~10%) overhead coming from applying the volume envelope to the sound. This could in principle be sped up by detecting the case of a completely flat envelope at the start and treating it as a special case, but I would think that it's not worth the effor...

You can speed things up by making sure the whole file is read into RAM: increase the "Maximal RAM size" in the advanced preferences to something like 8096. (That's assuming your computer has at least 16GB of RAM, otherwise you may end up swapping, which is actually *less* efficient.) With ...

Yes, I received it and was able to reproduce the problem, thanks!

Just send me a WeTransfer / Dropbox / iCloud link to martin@hairersoft.com.

Chances are that some of that sound data is being stored on disk, so it's a simultaneous read / write rather than a pure write. Also, even though WAV files aren't compressed, the sound data has to be converted from its internal 32-bit floating point representation to the 16 / 24bit integer represent...

So my point is that AP should simply leave anything that it doesn't read and know as is. :) This would be a bad idea since the 'bext' chunk contains some information about the waveform (peak level, peak loudness, etc) as well as a unique material identifier (UMI) used to identify the file. Changing...

Sorry, I cannot reproduce this, both resampling methods give me identical sonograms: