As the title says 
I have here old tape recordings where the high-end frequencies are non existing, and a recent recording of a friend that is just the same (in that case the result of massive use of pedals and bad choices).
1 Like
Maybe AI can do that in the future. Just like they can now turn old and grainy black and white video material into color and higher definition video.
However, it is no guarantee that it will sound exactly like the original. AI will just guess how the original would sound like, based on all data is is fed by. For common music like pop and rock, it will probably work well, but for more experimental music it may sound very different.
1 Like
i hope itās close! would breathe life into so many old/lost recordings.
imagine it let loose cleanign & restoring the internet archives 78rpm shellac collectionā¦
but yes, it will bring its own colour and decisions but hopefully user controllable to some extent.
even the stuff that is starting to utilise this area of tech is really exciting.
2 Likes
Should be possible already. Doesnāt sound like something modern neural networks arenāt capable of.
You can do it yourself if you have a decent computer, a suitable dataset (lots of pairs of originals and tracks recorded on tape) and lots of free time.
1 Like
I would try to clean some of the noise manually with tool like Izotopās Iris 2 and then mix together a duplicate track, cutting the low, and use an eq with frequency shifting. Maybe compressor as a final fxā¦
Btw, audicity has a very simple and smart noise remover that worth trying in situation like this. Good luck
1 Like
Celemony | Capstan works for tape, but it sounds as if youād be better possibly de-āmudā-ing and removing messier mid-frequencies?
Depends on what youāre training the materials on, i sāpose.
@roboKob it wasnāt anything serious, just that Jeanne clarified there was no data to filter, thereās nothing there, so I felt that my comment about your comment was no longer relevant. Iām not aware of a way to synthesize top end data that doesnāt exist beyond painful attempts to recreate individual sounds through hard synthesis, but I think some people may be on the right track thinking that AI could do it as a batch production type of work.
anyways sorry for the mystery!
And regarding room mics, I tried to record wind instruments with a room mic and ended up with a shit ton of noise that I didnāt want, but when I used the audacity tool it killed the sound of the instrument. Thatās what my post was specifically regarding.
2 Likes
I think I need to clarify that in the cases I mean, there is nothing to extract because ā¦ there is nothing Iāll edit my post.
2 Likes
What were the highs that got cut in context? Were they the overtones of existing content in the mids? If so, you might be able to sweeten it up a little running it through a harmonizer with low cut/high boost? Or even, like, a long tail shimmer with dampened low mixed in very lightly ā maybe with slow-release comp for a gate?
2 Likes
You could have a look at zynaptiq unfilter. It might be able to bring back some lost information.
2 Likes
@shigginpit and @Jeanne you both deleted your replay, my first ever web-embracing moment, and curious 
by the way, what I wrote tried and had success with noise and stuff from the room mics.
I hadnāt heard of this. But man, from the videos it sounds magic!
2 Likes
The spectral recovery function in RX is designed to recreate frequencies above 4K - might be worth a shot. Iāve not used it though
1 Like
Uuuuuuuuuuh!!! āSpectral Recovery is particularly useful in handling VoIP recordings, such as from Skype or Zoom, where there is a hard cutoff above which no audio exists.ā
jeanne, when elated, has turned into master p.
hopefully this yields a nice result, let us know how it works out.
1 Like
Itās intended for podcasts rather than music, but you might want to check out Adobe Enhance as well (itās free).
Some of the examples Iāve heard from that have been very impressive.
Gullfoss does a great job of boosting hi-end in a tasteful manner. The trial is free for 14 days, give it a go.
Just found this while looking into cheaper alternatives ā¦ this product has something called āVitalizeā which is supposed to do the same, but this Suite is $60, while the Izotope Suite is $1200.
āVitalize doesnāt boost certain frequencies, but adds harmonics instead. These harmonics are not based on distortion, but are artificially created by a sophisticated algorithm.ā
I actually use some Acon stuff already, so I might get this one
3 Likes
uh oh. look out syntakt, thereās a new falsetto in town.