teticio / deej-ai Goto Github PK

When trying to install requirements on Windows 10 via ' pip install -r requirements.txt' I receive the following error:

ERROR: Could not find a version that satisfies the requirement tensorflow==2.2.0 (from versions: 2.5.0, 2.5.1, 2.5.2, 2.5.3, 2.6.0rc0, 2.6.0rc1, 2.6.0rc2, 2.6.0, 2.6.1, 2.6.2, 2.6.3, 2.6.4, 2.6.5, 2.7.0rc0, 2.7.0rc1, 2.7.0, 2.7.1, 2.7.2, 2.7.3, 2.8.0rc0, 2.8.0rc1, 2.8.0, 2.8.1, 2.8.2, 2.9.0rc0, 2.9.0rc1, 2.9.0rc2, 2.9.0, 2.9.1)
ERROR: No matching distribution found for tensorflow==2.2.0

I attempted to install a later version of tensorflow which requires later versions of other packages but run into many errors later on.

I have ffmpeg, gstreamer, librosa, soundfile, audioread, yet none of my mp3s are processed.
Running Arch Linux, installed dependencies via pacman (some from AUR)

How do I get this working

Hi!

I've been interested in this project for quite some time now, and I finally have enough free time to look into it properly.
It's great that this is so simple to use with Spotify, however, my use case is slightly different. I would like to integrate this with my Jellyfin server to get suggestions for my local music. As far as I can tell, you've split up the different processing steps in a way that should allow analyzing local tracks instead of downloaded previews, but I think I need to get my Jellyfin track IDs in there somehow, instead of the Spotify IDs. Exporting to m3u would work fine for the start.

I also found the "Try it out for yourself" section in the readme, and that seems to be mostly working, but I believe that is using outdated code, right? The rewritten version only works for Spotify I'd guess.

Could you give me pointers on how and where to ideally modify the code, if necessary? Or is there a way to hook into the process, e.g. by generating some supplementary csv files based on my Jellyfin library?
Thanks in advance!

Starting with playlist 4920

No response or error after that, i've tried getting a new client id and a new playlist but now its at
Starting with playlist 0

That concludes spottily banned me for using this script.

I was able to install all the packages on Windows, no issues (although pandas was a bit pain in the neck, but managed to install the specified version), all pip packages are resolved...
But when I try to run the command python MP3ToVec.py Pickles mp3tovec --scan D:\Music I get following error:

  File "MP3ToVec.py", line 42, in <module>
    model = load_model(model_file)
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\saving\save.py", line 184, in load_model
    return hdf5_format.load_model_from_hdf5(filepath, custom_objects, compile)
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\saving\hdf5_format.py", line 194, in load_model_from_hdf5
    training_config, custom_objects))
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\saving\saving_utils.py", line 215, in compile_args_from_training_config
    loss = _deserialize_nested_config(losses.deserialize, loss_config)
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\saving\saving_utils.py", line 255, in _deserialize_nested_config
    return deserialize_fn(config)
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\losses.py", line 1835, in deserialize
    printable_module_name='loss function')
  File "C:\Users\Carlos\Anaconda3\envs\deejai\lib\site-packages\tensorflow\python\keras\utils\generic_utils.py", line 392, in deserialize_keras_object
    raise ValueError('Unknown ' + printable_module_name + ':' + object_name)
ValueError: Unknown loss function:cosine_proximity

I searched a bit here and there and found something promising
It says to modify the load_model and make it look like load_model(model_file, compile=False), so I did that, but I got yet another error this time:

Traceback (most recent call last):
  File "MP3ToVec.py", line 48, in <module>
    n_mels     = model.layers[0].input_shape[1]
IndexError: list index out of range

Apparently, the model.layers[0].input_shape is a list of only one element (instead of at least 3, the next line was accessing input_shape[2].
At this point I am uncertain what it is I am doing wrong, any ideas? Thanks!

Hello, and thanks for the good work :)

I have an issue when I try to generate a playlist, I tried with various input songs and get the same error.

Thanks in advance :)

 python Deej-A.I.py Pickles mp3tovec --playlist playlist_outfile.m3u --inputsong max-sharks/Soul\ Edifice\ -\ National\ Insurrection.mp3
26 MP3s
Outfile playlist: playlist_outfile.m3u
Input song selected: max-sharks/Soul Edifice - National Insurrection.mp3
Requested None songs
Traceback (most recent call last):
  File "Deej-A.I.py", line 488, in <module>
    tracks = make_playlist([input_song], size=n_songs + 1, noise=noise, lookback=lookback)
  File "Deej-A.I.py", line 208, in make_playlist
    similar = most_similar(positive=playlist[-lookback:], topn=max_tries, noise=noise)
  File "Deej-A.I.py", line 179, in most_similar
    mp3_vec_i = np.sum([mp3tovec[i] for i in positive] + [-mp3tovec[i] for i in negative], axis=0)
  File "Deej-A.I.py", line 179, in <listcomp>
    mp3_vec_i = np.sum([mp3tovec[i] for i in positive] + [-mp3tovec[i] for i in negative], axis=0)
KeyError: 'max-sharks/Soul Edifice - National Insurrection.mp3'

Quality of generated playlists is really great, thank you for your work!
Along the way, it allows to deduplicate the music library, since the same tracks follow each other)) I've been dreaming of doing something like this for a very long time!
Do you consider downloading texts and searching for the nearest ones not only by spectrogram, but also by "meaning" as the development of the project? Perhaps a linear combination of similarity in music and text...This could allow you to search for at least tracks in one language)

Hello! I've just discovered this and I'm trying it!
I set up an Ubuntu LXC container, installed all the things included the requirements.txt, and launched the scan process with success. Now I have my Picker directory with all the data.

python3 MP3ToVec.py Pickles mp3tovec --scan /home/ubuntu/deejai/ayj/

I'm using a small set of files (52 mp3) to test this. Then I created a tracks.txt file, containing just

'06 - Ella Fitzgerald - Cheek To Cheek.mp3'
'07 - Michael Buble - Come Fly With Me.mp3'
'08 - Paolo Nutini - Coming Up Easy.mp3'
'09 - Karen Souza - Creep.mp3'

or

'/home/ubuntu/deejai/ayj/06 - Ella Fitzgerald - Cheek To Cheek.mp3'
'/home/ubuntu/deejai/ayj/07 - Michael Buble - Come Fly With Me.mp3'
'/home/ubuntu/deejai/ayj/08 - Paolo Nutini - Coming Up Easy.mp3'
'/home/ubuntu/deejai/ayj/09 - Karen Souza - Creep.mp3'

But when I try

python3 Join_the_dots.py Pickles/mp3tovecs/mp3tovec.p --input tracks.txt mix.mp3 6

I always get the KeyError

Traceback (most recent call last):
  File "/home/ubuntu/deejai/Deej-AI/Join_the_dots.py", line 129, in <module>
    playlist = join_the_dots(input_tracks, n=n, noise=noise)
  File "/home/ubuntu/deejai/Deej-AI/Join_the_dots.py", line 67, in join_the_dots
    start_vec = mp3tovec[start]
KeyError: "'06 - Ella Fitzgerald - Cheek To Cheek.mp3'"

What am I doing wrong?

Trying to do python3 Deej-A.I.py Pickles mp3tovec or trying to create the playlist gives me this error:

Traceback (most recent call last):
  File "/home/ubuntu/deejai/Deej-AI/Deej-A.I.py", line 32, in <module>
    app = dash.Dash()
  File "/home/ubuntu/.local/lib/python3.10/site-packages/dash/dash.py", line 236, in __init__
    self.server.before_first_request(self._setup_server)
AttributeError: 'Flask' object has no attribute 'before_first_request'. Did you mean: '_got_first_request'?

Just to add some more info.

Not really sure what's up, I tried to run the code with the requirements (it couldn't find the right version of tensorflow so I just installed the latest one) and then this happens when I try to scan my music library.

C:\Users\Panda\Documents\Codestuffs\Deej-AI-master>python MP3ToVec.py Pickles mp3tovec --scan "E:\Panda\Music\iTunes\iTunes Media\Music"
2022-11-14 23:03:32.790183: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
2022-11-14 23:03:32.790836: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Creating Track2Vec matrices
2022-11-14 23:03:35.199784: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library nvcuda.dll
2022-11-14 23:03:35.219181: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties:
pciBusID: 0000:01:00.0 name: NVIDIA GeForce RTX 3080 computeCapability: 8.6
coreClock: 1.785GHz coreCount: 70 deviceMemorySize: 12.00GiB deviceMemoryBandwidth: 849.46GiB/s
2022-11-14 23:03:35.219800: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
2022-11-14 23:03:35.220358: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cublas64_11.dll'; dlerror: cublas64_11.dll not found
2022-11-14 23:03:35.220845: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cublasLt64_11.dll'; dlerror: cublasLt64_11.dll not found
2022-11-14 23:03:35.221412: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cufft64_10.dll'; dlerror: cufft64_10.dll not found
2022-11-14 23:03:35.222379: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'curand64_10.dll'; dlerror: curand64_10.dll not found
2022-11-14 23:03:35.222909: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cusolver64_11.dll'; dlerror: cusolver64_11.dll not found
2022-11-14 23:03:35.223424: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cusparse64_11.dll'; dlerror: cusparse64_11.dll not found
2022-11-14 23:03:35.223939: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudnn64_8.dll'; dlerror: cudnn64_8.dll not found
2022-11-14 23:03:35.224073: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1766] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2022-11-14 23:03:35.224545: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX AVX2
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2022-11-14 23:03:35.225276: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1258] Device interconnect StreamExecutor with strength 1 edge matrix:
2022-11-14 23:03:35.225335: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1264]
Traceback (most recent call last):
File "C:\Users\Panda\Documents\Codestuffs\Deej-AI-master\MP3ToVec.py", line 42, in
model = load_model(model_file)
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\saving\save.py", line 201, in load_model
return hdf5_format.load_model_from_hdf5(filepath, custom_objects,
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\saving\hdf5_format.py", line 198, in load_model_from_hdf5
model.compile(**saving_utils.compile_args_from_training_config(
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\saving\saving_utils.py", line 212, in compile_args_from_training_config
loss = _deserialize_nested_config(losses.deserialize, loss_config)
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\saving\saving_utils.py", line 253, in _deserialize_nested_config
return deserialize_fn(config)
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\losses.py", line 2020, in deserialize
return deserialize_keras_object(
File "C:\Users\Panda\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\utils\generic_utils.py", line 698, in deserialize_keras_object
raise ValueError(
ValueError: Unknown loss function: cosine_proximity. Please ensure this object is passed to the custom_objects argument. See https://www.tensorflow.org/guide/keras/save_and_serialize#registering_the_custom_object for details.

Hello, just found this project and set it up. However, when accessing the webpage to start, the sliders for Keep On and Drunk are not accessible to those who use a screenreader. I use nvda as my screenreader, and the only things that I can see are several clickable objects and the numbers for the start and end of the sliders, but no way to adjust them. When using object navigation, I see the NVDA says Diagram where I think the sliders should be.

Hi,

Great work with this. I have been building a bit on the side of this for about a year on some ideas, but then I noticed the terms of service for development in Spotify API, do you have any better knowledge around this?
https://developer.spotify.com/terms#section-iv-restrictions

General restrictions.

Misuse of the Spotify Platform. Do not misuse the Spotify Platform, including by
    using the Spotify Platform or any Spotify Content to train a machine learning or AI model or otherwise ingesting Spotify Content into a machine learning or AI model;

10. Governing Law

Notwithstanding any other provision in the Developer Agreement, this Appendix shall be governed by, and interpreted in accordance with, the laws of Sweden.

Cheers
dasbts

Great App! I was trying to make it work with one of my Spotify playlists and was unable to understand how tokenization works with the redirect URI...could you add instructions to either the colab page or the readme.md? Thanks so much!

I can't seem to figure out why?

last 10 lines:

Skipping /Users/dariuspleasant/Documents/Music/Music 2/79rs Gang/Expect The Unexpected/07 79rs Gang and Lakou Mizik- Iko Kreyòl (79rs Version).mp3
100%|█████████████████████████████████████| 1362/1362 [00:20<00:00, 65.92file/s]
Creating MP3ToVecs for batch 1/1
Precalculating cosine distances
0vector [00:00, ?vector/s]
Calculating IDF weights
0vector [00:00, ?vector/s]
Calculating TF weights
0mp3 [00:00, ?mp3/s]

pip freeze

absl-py==0.13.0
astunparse==1.6.3
audioread==2.1.9
Brotli==1.0.9
cachetools==4.2.2
certifi==2021.5.30
charset-normalizer==2.0.3
click==8.0.1
dash==0.42.0
dash-core-components==0.47.0
dash-daq==0.1.4
dash-html-components==0.16.0
dash-renderer==0.23.0
dash-table==3.6.0
dataclasses==0.8
decorator==5.0.9
Flask==1.0.2
Flask-Compress==1.10.1
gast==0.3.3
google-auth==1.33.1
google-auth-oauthlib==0.4.4
google-pasta==0.2.0
grpcio==1.39.0
h5py==2.10.0
idna==3.2
importlib-metadata==4.6.1
itsdangerous==2.0.1
Jinja2==3.0.1
joblib==1.0.1
Keras==2.3.1
Keras-Applications==1.0.8
Keras-Preprocessing==1.1.2
librosa==0.6.3
llvmlite==0.31.0
Markdown==3.3.4
MarkupSafe==2.0.1
mutagen==1.42.0
numba==0.48.0
numpy==1.16.3
oauthlib==3.1.1
opt-einsum==3.3.0
pandas==0.24.2
Pillow==8.3.1
plotly==5.1.0
protobuf==3.17.3
pyasn1==0.4.8
pyasn1-modules==0.2.8
python-dateutil==2.8.2
pytz==2021.1
PyYAML==5.4.1
requests==2.26.0
requests-oauthlib==1.3.0
resampy==0.2.2
rsa==4.7.2
scikit-learn==0.24.2
scipy==1.4.1
six==1.16.0
spotipy==2.4.4
tenacity==8.0.1
tensorboard==2.2.2
tensorboard-plugin-wit==1.8.0
tensorflow==2.2.0
tensorflow-estimator==2.2.0
termcolor==1.1.0
threadpoolctl==2.2.0
tqdm==4.31.1
typing-extensions==3.10.0.0
urllib3==1.26.6
Werkzeug==2.0.1
wrapt==1.12.1
zipp==3.5.0

Often I get

Traceback (most recent call last):
  File "f:\Stuff\Programming\libs\deejAI.py", line 6, in <module>
    audio_encoder.encode(files)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\audiodiffusion\audio_encoder.py", line 89, in encode
    self.mel.load_audio(audio_file)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\diffusers\pipelines\audio_diffusion\mel.py", line 98, in load_audio
    self.audio, _ = librosa.load(audio_file, mono=True, sr=self.sr)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\librosa\util\decorators.py", line 88, in inner_f
    return f(*args, **kwargs)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\librosa\core\audio.py", line 218, in load
    y = resample(y, orig_sr=sr_native, target_sr=sr, res_type=res_type)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\librosa\util\decorators.py", line 88, in inner_f
    return f(*args, **kwargs)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\librosa\core\audio.py", line 686, in resample
    y_hat = resampy.resample(y, orig_sr, target_sr, filter=res_type, axis=-1)
  File "D:\Mambaforge\envs\audiodiffusion\lib\site-packages\resampy\core.py", line 117, in resample
    raise ValueError(
ValueError: Input signal length=0 is too small to resample from 44100->22050

and it stops the whole program
example audiofile:
100 gecs - Hand Crushed By A Mallet (Soup Remix).zip

the issue is that when librosa opens it it has shape 2,0

if anyone has the same issue (i have it with every 20th mp3 approximately)
go to D:\Mambaforge\envs\audiodiffusion\lib\site-packages\diffusers\pipelines\audio_diffusion\mel.py
search for def load_audio
and replace
if audio_file is not None:
self.audio, _ = librosa.load(audio_file, mono=True, sr=self.sr)

with
if audio_file is not None:
try:
self.audio, _ = librosa.load(audio_file, mono=True, sr=self.sr)
except:
try:
import madmom
self.audio, _ = madmom.io.audio.load_audio_file(audio_file, dtype=float, sample_rate=self.sr, num_channels=1)
self.audio=self.audio.T
except:
try:
import pedalboard.io
with pedalboard.io.AudioFile(audio_file).resampled_to(self.sr) as f:
self.audio = f.read(f.frames)
#samplerate = f.samplerate
except:
import soundfile
self.audio, _ = soundfile.read(audio_file, samplerate=self.sr, channels=1)
self.audio=self.audio.T

also install ffmpeg-python and pedalboard

what python version do i need???

Hi, really cool project!
I have 2 more suggestions for data sources to train the model on.
https://last.fm and https://listenbrainz.org/ allow users to track the songs they are listening to (and allow you to query that information). While you don't get playlists, the plays are tagged with a timestamp, so you could look for consecutive plays of songs and split e.g. when there is a break of more than 30min between songs.

Hi teticio,

I find your work really impressive, congrats on this great work!
Using it I though that the ability to export the generated playlist to a .m3u file would be a great feature to make the program more portable.
Have you had any thoughts on that?

Hi , got a bug :
Generated database with command (20 of 20 has scanned)
python MP3ToVec.py Pickles mp3tovec --scan C:\mp3library
Than generated m3u with
python Deej-A.I.py Pickles mp3tovec --playlist C:\m3ufiles\000000018.m3u8 --inputsong C:\mp3library\66.mp3 --nsongs 3
Got correct m3u ( with similar songs)
Next added another song to C:\mp3library , and ran the command again
python MP3ToVec.py Pickles mp3tovec --scan C:\mp3library
New song added to database (1 of 1 has scanned)

When tried to get m3u again
python Deej-A.I.py Pickles mp3tovec --playlist C:\m3ufiles\000000018.m3u8 --inputsong C:\mp3library\66.mp3 --nsongs 3
got error :
C:\Deej-AI-master\Deej-A.I.py:186: RuntimeWarning: invalid value encountered in scalar divide
cos_proximity = np.dot(mp3_vec_i, mp3_vec_j) / (np.linalg.norm(mp3_vec_i) * np.linalg.norm(mp3_vec_j))
Got m3u generated , but songs in it not similar at all

In short - after adding new song to lib and additional scan , that leads to error (with wrong(not similar) m3u output)

Hi there. Most of my songs are in flac format.
It'd be nice to have flac support.
Is this doable?

I know it was mentioned in there, but how would you go about linking a Spotify to this so it can recommend songs on Spotify?

I have some trouble with the requirements on my MacBook with M2 processor. I modify the requirements.

Add a separate requirement file for MacOS:

dash
tqdm
mutagen
Flask
spotipy
librosa
dash_core_components==0.47.0
dash_html_components==0.16.0
dash_daq==0.1.4
pandas
numpy
dash_html_components
Pillow
tensorflow-macos
numba

Reference: https://developer.apple.com/metal/tensorflow-plugin/