How can I run this project to generate MIDI

[dedededefo]

Is it according to demo.Ipynb steps?

[netpi]

@dedededefo You can run this colab to train a model and generate midi. Or use this Trained Model | Step28500 with parameters below:

emb_sizes = [64, 256, 768, 256, 512, 256, 256, 256]
config = {
    "vocab_sizes": vocab_sizes,
    "emb_sizes": emb_sizes,
    "d_model": 512,
    "dff": 1024,
    "num_layers": 12,
    "num_heads": 4,
    "dropout_rate": 0.1,
    "length": 1024,
    "rpr": True,
    "dataset": f'{database_name}_{train_seq_length}',
}

[dedededefo]

@dedededefo You can run this colab to train a model and generate midi. Or use this Trained Model | Step28500 with parameters below:

emb_sizes = [64, 256, 768, 256, 512, 256, 256, 256]
config = {
    "vocab_sizes": vocab_sizes,
    "emb_sizes": emb_sizes,
    "d_model": 512,
    "dff": 1024,
    "num_layers": 12,
    "num_heads": 4,
    "dropout_rate": 0.1,
    "length": 1024,
    "rpr": True,
    "dataset": f'{database_name}_{train_seq_length}',
}

Sorry to bother you again. I downloaded the pre training model, but I'm not sure how to use it. Do I need the file path TF log? I tried to run the generation according to collab, but it seems that I can't use the pre training model

[netpi]

@dedededefo a new Colab to generate MIDI with step28500

https://colab.research.google.com/drive/1-r5IpQgOoZgkOGof8EdjBkQzb6H6e64J?usp=sharing

[LqNoob]

@dedededefo a new Colab to generate MIDI with step28500

https://colab.research.google.com/drive/1-r5IpQgOoZgkOGof8EdjBkQzb6H6e64J?usp=sharing

Hi, @netpi . I have a question about music generation is that how to generate audible audio files in other formats after generating MIDI files because MIDI files are not a common audio format.

[netpi]

@dedededefo a new Colab to generate MIDI with step28500 https://colab.research.google.com/drive/1-r5IpQgOoZgkOGof8EdjBkQzb6H6e64J?usp=sharing

Hi, @netpi . I have a question about music generation is that how to generate audible audio files in other formats after generating MIDI files because MIDI files are not a common audio format.

hi @LqNoob Generating raw audio(Singing) is another field (SVS). You can search Singing Voice Synthesis to see more. Or you can use magenta.js if you just want to show midis on websites.

[LqNoob]

Thanks for your kind reply. @netpi