Chinwonder2
Global Moderator 
RIP QueenZone 1995 - 2020
Posts: 588 
Likes: 904
|
Post by Chinwonder2 on Oct 20, 2021 21:04:09 GMT
|
|
BrƎИsꓘi
Administrator
They called it paradise, I don't know why...You call some place paradise, kiss it goodbye.
Posts: 3,740
Likes: 2,873
|
Post by BrƎИsꓘi on Oct 20, 2021 21:16:46 GMT
that's impressive. does it work as a standalone app? or is it web-based?
|
|
Chinwonder2
Global Moderator
RIP QueenZone 1995 - 2020
Posts: 588
Likes: 904
|
Post by Chinwonder2 on Oct 20, 2021 21:17:27 GMT
It's all web-based as far as I know  -Chin |
|
Steve
Wordles & Heardles 
Queen Mab
😀
Posts: 4,220
Likes: 942
|
Post by Steve on Oct 20, 2021 22:28:42 GMT
This sounds awesome! Cheers mate.
|
|
|
|
Post by Dry Paint Dealer Undr on Oct 21, 2021 10:56:54 GMT
|
|
Golden Salmon
Wordles & Heardles
Politician
Posts: 827
Likes: 817
|
Post by Golden Salmon on Oct 21, 2021 16:30:46 GMT
Wow, that's the closest Innuendo instrumental yet! Thank you for it.
Other than updates to these AI tools, perhaps some other songs will yield better results given their features. There's this recent "Don't Try So Hard" fake instrumental that's nearly convincing.
|
|
qsmjohn
Dragonfly Trumpeter
Posts: 112
Likes: 59
|
Post by qsmjohn on Oct 21, 2021 22:12:09 GMT
Amazing! So far the best instrumental of Innuendo imo. Do you think that in the future it will become more and more difficult to distinguish fake instrumentals and real instrumentals? I think there will be always a hearable limit, but who knows...
|
|
|
|
Post by queenvault on Oct 22, 2021 0:52:23 GMT
How does the stand-alone vocal hold up?
|
|
|
|
Post by pimderks on Oct 22, 2021 10:17:39 GMT
Ho-ly s*** that's impressive. Anyone planning to pull the entire discography through this AI?  |
|
Chinwonder2
Global Moderator
RIP QueenZone 1995 - 2020
Posts: 588
Likes: 904
|
Post by Chinwonder2 on Oct 22, 2021 10:49:14 GMT
How does the stand-alone vocal hold up? I'll send it to you via PM later on, at University right now lol Ho-ly s*** that's impressive. Anyone planning to pull the entire discography through this AI? I plan to do this to most of the videos I'll be uploading to my channel, eventually. Creating an instrumental/karaoke version for each song where one doesn't currently exist. -Chin |
|
DasTarD
Satyr
www.DasTarD.nl
Posts: 78
Likes: 48
|
Post by DasTarD on Oct 22, 2021 12:03:19 GMT
I have started with White Queen, it works great :-D
|
|
|
|
Post by Brian's Wig on Oct 22, 2021 13:47:35 GMT
I've already done all the albums using online software about 3 months ago (and I mean ALL the albums, solo as well), but the new rules here say I can't share them.
Certainly haven't the time or patience to piss about putting them on youtube. Sorry guys.
There's Lalal.Ai which has to be paid for or there's X-Minus which actually gives better results, but sound slightly muddier.
BUT they all have an issue with the overdubs and some of Brian's high guitar work.
|
|
georg
Global Moderator
wrote several books
Posts: 1,055
Likes: 1,358
|
Post by georg on Oct 22, 2021 15:12:44 GMT
I've already done all the albums using online software about 3 months ago (and I mean ALL the albums, solo as well), but the new rules here say I can't share them. Certainly haven't the time or patience to piss about putting them on youtube. Sorry guys. There's Lalal.Ai which has to be paid for or there's X-Minus which actually gives better results, but sound slightly muddier. BUT they all have an issue with the overdubs and some of Brian's high guitar work. Technically all you'd need to do would be post a sample of one song on YT and then if there so happens to be a link in the description to another site then that's just the way the cookie crumbles, but you didn't hear that from me... |
|
kosimodo
Dragonfly Trumpeter
Posts: 158
Likes: 79
|
Post by kosimodo on Oct 22, 2021 17:59:41 GMT
^^ a wise man said..
|
|
DasTarD
Satyr
www.DasTarD.nl
Posts: 78
Likes: 48
|
Post by DasTarD on Oct 22, 2021 18:05:19 GMT
I've already done all the albums using online software about 3 months ago (and I mean ALL the albums, solo as well), but the new rules here say I can't share them. Certainly haven't the time or patience to piss about putting them on youtube. Sorry guys. There's Lalal.Ai which has to be paid for or there's X-Minus which actually gives better results, but sound slightly muddier. BUT they all have an issue with the overdubs and some of Brian's high guitar work. Was that done with Spleeter? This new software seems to work better, at least for me. |
|
rub3945
Ploughman
Posts: 252
Likes: 437
|
Post by rub3945 on Oct 23, 2021 0:29:55 GMT
Excellent! Thanks Chinwonder2 and Dry Paint Dealer Undr for sharing. Tried this out with a few songs so far and the results are great! Certain songs have turned out better than others based on their complexities. TMOTBQ for example didn't turn out perfectly. But others like It's Late are damn near perfect so far! Bits and pieces of Brian's guitar solos are sometimes picked up but shouldn't be too hard to patch up with the originals if necessary. I'll share a few of the results here (instrumental and vocals). I probably shouldn't be posting them here directly as per the rules but will follow the wise man's advice and share something shortly 😉 |
|
rub3945
Ploughman
Posts: 252
Likes: 437
|
Post by rub3945 on Oct 23, 2021 1:07:38 GMT
There we are. If you dare click through to the YouTube page from here, legend says the description of the video may contain some hidden treasures 🤫
|
|
|
|
Post by dragonkiller on Oct 23, 2021 11:41:30 GMT
I dared to click and found some amazing stuff...thank you and thanks to Chin for starting it off.
|
|
onedunpark
Wordles & Heardles
Apothecary Man
Posts: 8,299
Likes: 2,588
|
Post by onedunpark on Oct 26, 2021 7:33:50 GMT
I clicked through as well. This is great, the AI has done an incredible job. Many thanks to all involved.
|
|
|
|
Post by fireplace on Oct 26, 2021 15:28:42 GMT
Clicking through quite happily 😊. Thanks for posting!
|
|
scandal
Tatterdemalion
Posts: 24
Likes: 33
|
Post by scandal on Oct 26, 2021 18:20:13 GMT
There we are. If you dare click through to the YouTube page from here, legend says the description of the video may contain some hidden treasures 🤫 "Mad the Swine" acapella sounds really great bringing all these vocal layers upfront. Could anyone pls do the same with "Need Your Loving Tonight"? May be it would be possible to make those barely audible vocals, which are scattered all around the song and presumably leaked from guiding track, clearly heard? |
|
onedunpark
Wordles & Heardles
Apothecary Man
Posts: 8,299
Likes: 2,588
|
Post by onedunpark on Oct 27, 2021 10:55:01 GMT
The start of the Pain Is So Close to Pleasure Instrumental is so close to the start of The Way You Make Me Feel by Michael Jackson (released the year afterwards) , I had to do a double-take, when it came on. Maybe it's just generic disco beat stuff, and I don't listen to a lot of that, but it made me smile if nothing else. Loving this AI stuff. |
|
Dimitris
Ploughman
Posts: 476
Likes: 282
|
Post by Dimitris on Oct 27, 2021 12:03:25 GMT
The start of the Pain Is So Close to Pleasure Instrumental is so close to the start of The Way You Make Me Feel by Michael Jackson (released the year afterwards) , I had to do a double-take, when it came on. Maybe it's just generic disco beat stuff, and I don't listen to a lot of that, but it made me smile if nothing else. Loving this AI stuff. On the 12inch remix. |
|
pg
Queen Mab
Posts: 2,082
Likes: 1,348
|
Post by pg on Nov 4, 2021 8:54:46 GMT
Isn't this the solution to "there's no isolated piano mike from Earl's Court"?
|
|
pg
Queen Mab
Posts: 2,082
Likes: 1,348
|
Post by pg on Nov 4, 2021 8:59:59 GMT
I'm new to Colab - where does the output go? It's processed my file but can't see any output files, even in the folder with "Processed Outputs go here.txt" in it |
|
Egietje
Wordles & Heardles
Politician
Posts: 578
Likes: 486
|
Post by Egietje on Nov 4, 2021 12:22:53 GMT
I'm new to Colab - where does the output go? It's processed my file but can't see any output files, even in the folder with "Processed Outputs go here.txt" in it They should go into VocalRemover5-COLAB/separated in your Google Drive |
|
BrƎИsꓘi
Administrator
They called it paradise, I don't know why...You call some place paradise, kiss it goodbye.
Posts: 3,740
Likes: 2,873
|
Post by BrƎИsꓘi on Nov 4, 2021 12:50:10 GMT
|
|
pg
Queen Mab
Posts: 2,082
Likes: 1,348
|
Post by pg on Nov 4, 2021 13:19:03 GMT
1. I'm no proof of success
2. I'm eligible for SAGA holidays and retirement complexes
3. You're technical enough when it comes to pictures....
|
|
BrƎИsꓘi
Administrator
They called it paradise, I don't know why...You call some place paradise, kiss it goodbye.
Posts: 3,740
Likes: 2,873
|
Post by BrƎИsꓘi on Nov 4, 2021 14:06:46 GMT
3. You're technical enough when it comes to pictures.... you'd think. but my brain doesn't agree. |
|
Egietje
Wordles & Heardles
Politician
Posts: 578
Likes: 486
|
Post by Egietje on Nov 5, 2021 1:30:21 GMT
First, go to the Colab: colab.research.google.com/drive/1eK4h-13SmbjwYPecW2-PdMoEbJcpqzDt?usp=sharingHere, press the play button next to Mount to Drive, this will run for a bit and then prompt you to go to a link.  This link will give the colab full access to your Drive files, so it might be wise to do this on a separate account as to not put your files at risk. When at the link, sign in with the right account (the one you also opened the Colab with). It will give you a code, copy this and paste it into the prompt back at the Colab. It should now start installing the software/running updates, after which it should say Success! In case errors arise, it should be fine after running the Mount to Drive script again, although I have not run into problems.  Next, upload any tracks you want to separate into the VocalRemover5-COLAB/tracks folder on your Drive. These can be any music codec you want (WAV, FLAC, MP3, etc.).  When these have been uploaded, run the Print a list of tracks and models script (by again using the play button next to it). This will print all of the music files detected, as well as a list of AI models which can be ignored.  Finally, we get to actually separating the files. Before running the script, we need to configure some settings. First off all, check convertAll, as this will simply convert all of the detected files, rather than having to specify one. Then, check tta, this will improve the quality of the separation. All other settings can be left as is.  Once configured, run the Inference script by using the play button. It will now start separating all files, this might take a while. Luckily, a progress bar is included in the output at the bottom. You might run into the error where it says 'all CUDA-capable devices are busy or unavailable', or something similar.  This means currently, all GPUs used for the process are unavailable. To fix this, uncheck gpu from the settings. This does mean the separation will be way slower, about 15x, but there is no other way I've found sadly.  After this, run the Inference script again. When separation has been completed, all separated tracks should now appear in your Drive in the VocalRemover5-COLAB/separated folder. There will be 2 files for each uploaded track, the instrumental and the vocal track. These can either be downloaded individually, by right-clicking a file and clicking download, or all together by clicking on the name of the separated folder and pressing download here.  The files will be in WAV format. Hope this was helpful and you enjoy the tool! |
|
