I know version 11 has just been released with some great additions for Post-Production (Modules Panel / Modules Chain / Batch Processing), but there are still some essential tools missing for Post-Production, which I hope to see in the future:
De-Wind: Remove unwanted wind noise of field recordings
De-Rustle: Remove cloth noise of Lavalier microphones
De-Plosive: Remove pops of dialogue
With RX’s disastrous and embarrassing implementation of ARA2 support (only 3! of its more than 25 modules will be available?!?) there is now room for SpectraLayers to take over the throne of audio restoration. Therefore I hope to see more modules tailored towards Post-Production in the future.
More Post-Production features would be awesome. I’m also looking for an RX alternative in Post-Pro.
Unfortunately, I’m getting the impression that now SL is focused on modules related to unmixing music tracks.
Don’t worry, I try to equally push SL in music and post-prod with each new version. For instance SL11 introduced Voice DeClip, Unmix Crowd Noise, an improved Voice DeNoise, and an improved Unmix Multiple Voices.
Requests duly noted
It’s really refreshing to see a developer so active and engaged on the forum. Great job @Robin_Lobel Good to hear more Post-Production features will come in the future. Looking forward to ditching RX in the future all together and solely work with SpectraLayers.
De-warble, or whatever you would call it. A plugin to address poorly recorded / encoded lower res files that tend to warble. Like bad mp3s or phone recordings. Happens far too often these days.
A module that generates both fundamental and overtones from a bandwidth limited source. This is related to the above. Once the artifacts of the former are removed any signal that is bandwidth limited needs new high-end content generated. Same with fundamental to get back the “body” of the voice.
The second one is pretty important because it allows us to process problematic audio much harder and then just regenerate what we’ve cut out. Rustling is a pretty common problem since once that’s out the high end tend to disappear as well. So we can “nuke” the dialog and then regenerate to fix the problem we created when we fixed another problem.
Voice cloning. I really think you need to look into this, pronto. The future I bet will be that we take noise prints of an existing person and then apply that voice to someone else. This would allow us to take good sections of dialog for training and then just re-record ourselves a better, clean take that we can simply clone.
Think docs, lifestyle and reality. They cut together an episode using a text editor and the NLE then follows to cut together actual footage. Pitch, intonation, intensity is all over the place in one single sentence at times. We could just re-record real quick and apply the clone and we’d be done.
Btw - a function like this would make other functions in some sense obsolete. If there is problem with all of the above - rustle, poor recording, poor editing - then a simple new recording and cloning would fix it all in one go.
Consider that this is surely the future and sooner or later this will make its way into NLE’s, and surely we want to stay ahead of that before we become obsolete…
I agree with everything esp the 3rd. But to use SP11 for Post some PRE needs urgent fixing.
Foe example, VST3 plugin setting are not saved when you save the module chain or VST3 chain…Says you set Waves F6 set to highpass and Waves Sibilance and save it as a preset settings by the name ‘ Wave’, and when you want to reapply the settings they seem to reset to default.
Or for example you want to use RX-declick, Mouth declick, deplosive followed by Waves Sibilance and Waves F6.and save it as ‘Pre-Process Chain’ preset in the module chain you loose the settings making the preset useless as you end up redo doin every settings everytime you want to use the preset.,
May be it is normal behaviour or a fix is already on its way in the next patch.
Well done to @Robin_Lobel for all his efforts in getting SLP11 released…
The (great sounding) feature requests posted here for future post-pro consideration, seem like they deserve Robin having access to examples of what you guys are actually dealing with in your very real-world scenarios, however long or short.
I can imagine this help focus ‘solution’ efforts from his end… i.e. not (just) leaving him with his ‘interpretation’ of what you specifically have in front of you… only if it’s safe/non-sensitive material, of course…
In reality the problematic clip was run through Apples FCP’s AI assisted module by exporting several passes and opening them up in SLP9. Then using the excellent spectral selection tools we manually refined and fine tuned the voice print, and in the end using the same tools extracted and merged the ambience back. We ended up using roughly 20 layers in SLP9.
It took us 2 days just to fix a few seconds of dialogue and the same in SPL11 took less than 10 minutes.
Disaster strikes you where you least expect. And here something like Voice Cloning would be a perfect candidate for rescue and restoration of the soundbite in question.
Hi Robin. Agreed: your efforts addressing issues here are nothing short of amazing. If only the world had more involved people like you in it but I digress.
I’d like to request a feature to correct R2R and cassette tape speed issues currently addressed by Celemony Capstan. For instance, some battery powered recorders changed recording speed progressively as the battery wore down during a long session.
As you likely know, Capstan resolves the problem by correcting a subtle pilot frequent frequency (400 Hz?) from drifting thereby ensuring consistent speed. However, I’m not certain whether cheap, voltage driven recorders of the era even used pilot signals. Then, of course, quality recorders might have had mechanical issues during one-off recordings i.e.
record a session
place tape in box without quality check
Might it be possible for SL to include a module to correct tape speed drift? Eg. A4 is 440Hz at session start and ramps down to G4 392Hz an hour later as the battery runs down.
@BJ_Dobbs
Just to say, I’m getting the hang of removing the RFI I was talking about. Essentially, after running Unmix Noisy Speech; then, I’m setting a Spectral Region as a guide and moving the unwanted fizzes to another layer using various selection tools. If you still want a sample I can try to upload one later…
I do have RFI all over the place…which shows up as a continuous freq band…which cleans out very well with UnMix Noisy Speech.
The fffft sounds are more on a carrier…or movement by the contributor might have disturbed the antenna…
Senn EW 100G is what we were using…and had seen some abuse…not the best wireless gear IME
these quick little blasts that happen randomly…and can be pretty wide band
I haven’t tried the SL11 DeHum Module yet…as my previous workflows, I seldom used dehum for this…or at least Acon and Sonnox didn’t help with this particular condition…great for music, tho
What is working quite well is
Unmix Noisy Speech
then unmix levels at -65dB on problem phrases
I might bring some harmonics to the Hi layer, too
then attenuate the Low layer
