What's new
By:
Filters
LiteRECORDS

Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • Guest, before your account can be reviewed you must click the activation link sent to your email account. Please ensure you check your junk folders.
    If you do not see the link after 24 hours please open a support ticket.

Powerful New Vocal Remover AI - Instructions

djtayz said:
Did you install all of these?

pip install Pillow
pip install tqdm==4.30.0
pip install librosa==0.6.3 (it shows you're missing this one)
pip install opencv-python
pip install numba==0.48.0
pip install SoundFile
pip install soundstretch
Click to expand...

Working now. Definitely does a much better job on instrumentals, just very faint vocal bleed in places. I ran UB40-Red Red Wine and there's just little bits here and there of louder vocal bleeds.
As far as acapella making I really don't think good studio quality vocals can be extracted until technology evolves to the point where removed frequencies can be replaced artificially somehow. I just don't see any other way to do it. Like some AI trained to remember the singers voice frequencies and emulating them somehow. That's the biggest problem with RX7, Phonicmind, Spleeter etc etc.
Has anyone tried running things through twice? I wonder if it muddies the result.
 
Last edited:
IS anyone having same issues with NEW VERSION on stop 1 in Google Colab???

Enter your authorization code:
··········
---------------------------------------------------------------------------
TIMEOUT Traceback (most recent call last)
<ipython-input-1-96637c339ad2> in <module>()
2
3 from google.colab import drive
----> 4 drive.mount('/content/drive')

4 frames
/usr/local/lib/python3.6/dist-packages/pexpect/expect.py in timeout(self, err)
142 exc = TIMEOUT(msg)
143 exc.__cause__ = None # in Python 3.x we can use "raise exc from None"
--> 144 raise exc
145
146 def errored(self):

TIMEOUT: <pexpect.popen_spawn.PopenSpawn object at 0x7f6fbc828c18>
searcher: searcher_re:
0: re.compile('google.colab.drive MOUNTED')
1: re.compile('root@096134cdc36c-85141bb86cc94a149a09de7da1bd21db: ')
2: re.compile('(Go to this URL in a browser: https://.*)$')
3: re.compile('Drive File Stream encountered a problem and has stopped')
4: re.compile('drive EXITED')
<pexpect.popen_spawn.PopenSpawn object at 0x7f6fbc828c18>
searcher: searcher_re:
0: re.compile('google.colab.drive MOUNTED')
1: re.compile('root@096134cdc36c-85141bb86cc94a149a09de7da1bd21db: ')
2: re.compile('(Go to this URL in a browser: https://.*)$')
3: re.compile('Drive File Stream encountered a problem and has stopped')
4: re.compile('drive EXITED')
 
@djtayz




Same thing is happening with demucs.

I also tried with a different account...
 
Last edited:
joaoasq said:
---
Click to expand...

When you play the first step, go to the link it gives you, choose your account and allow Colab to use your Drive. After that you will get a long code that you need to copy and paste to the Colab project and click enter.
 
Last edited:
INGSOC said:
Going to test my theory that if you run through the process twice it might improve results.
Click to expand...

It doesn't. Your only bet is to either run it through the stacked model or use another AI to clean it up (but it's not guaranteed it will work on every song). Most people use RX 7 (RX 8 just got released btw)
 
Yeah confirmed makes zero difference. Also interesting that using 320 mp3 of the same song sounded exact same as lossless wav test I did. Still recommend using lossless if available. What is the stacked model?
 
INGSOC said:
Yeah confirmed makes zero difference. Also interesting that using 320 mp3 of the same song sounded exact same as lossless wav test I did. Still recommend using lossless if available. What is the stacked model?
Click to expand...

It's trained on instrumentals made with the app to clean up some vocal residue
 
@anjok what does HP and NP mean for the models? I've been using HP with good results, but just curious what the main difference would be?
 
INGSOC said:
@anjok what does HP and NP mean for the models? I've been using HP with good results, but just curious what the main difference would be?
Click to expand...

It means higher pitch. That model will do better with higher pitch vocals and pop songs vs the other with is better at heavier tracks as well as mid to low range vocals. I will be releasing another update to those models as time allows, hopefully within the next 2 weeks.
 
the google cloud step 1 does not accept my code. it gives error

Go to this URL in a browser: https://accounts.google.com/o/oauth...om/auth/peopleapi.readonly&response_type=code

Enter your authorization code:
··········
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-5-96637c339ad2> in <module>()
2
3 from google.colab import drive
----> 4 drive.mount('/content/drive')

/usr/local/lib/python3.6/dist-packages/google/colab/drive.py in mount(mountpoint, force_remount, timeout_ms, use_metadata_server)
247 wrote_to_fifo = True
248 elif case == 5:
--> 249 raise ValueError('mount failed: invalid oauth code')
250 if not wrote_to_fifo:
251 with open(fifo, 'w') as fifo_file:

ValueError: mount failed: invalid oauth code
 
ferser1 said:
the google cloud step 1 does not accept my code. it gives error

Go to this URL in a browser: https://accounts.google.com/o/oauth...om/auth/peopleapi.readonly&response_type=code

Enter your authorization code:
··········
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-5-96637c339ad2> in <module>()
2
3 from google.colab import drive
----> 4 drive.mount('/content/drive')

/usr/local/lib/python3.6/dist-packages/google/colab/drive.py in mount(mountpoint, force_remount, timeout_ms, use_metadata_server)
247 wrote_to_fifo = True
248 elif case == 5:
--> 249 raise ValueError('mount failed: invalid oauth code')
250 if not wrote_to_fifo:
251 with open(fifo, 'w') as fifo_file:

ValueError: mount failed: invalid oauth code
Click to expand...

Click the link. Get code. Paste code. I would manually highlight and copy. There is an issue with the copy button on that page.
 
You must log in or register to reply here.