matthewawesome / aicomposer Goto Github PK

Errors setting up the environment:

ERROR: pynndescent 0.5.1 has requirement numba>=0.51.2, but you'll have numba 0.49.1 which is incompatible.
ERROR: note-seq 0.0.2 has requirement numba==0.48.0, but you'll have numba 0.49.1 which is incompatible.
ERROR: kfac 0.2.3 has requirement tensorflow-probability==0.8, but you'll have tensorflow-probability 0.12.1 which is incompatible.
ERROR: tensor2tensor 1.15.7 has requirement tensorflow-probability==0.7.0, but you'll have tensorflow-probability 0.12.1

All worked with your midi files, so I tried load known Bach and Beethoven midi files, they loaded but I then got at
Trio Mashup Time!:
IndexError Traceback (most recent call last)
in ()
8 song_one_trio = 1#@param {type:"integer"}
9 song_two_trio = 0#@param {type:"integer"}
---> 10 start_trio = all_trios[0][song_one_trio]
11 end_trio = all_trios[1][song_two_trio]

Is there a specific setup of the midi that is needed? any help appreciated

Hi there,
I have been trying to change the length of the music (mean sequence). I tried to change some parameters in interpolate function as shown below.

def interpolate(model, start_seq, end_seq, num_steps, max_length=40,
                assert_same_length=True, temperature=0.5,
                individual_duration=5.0):

When i changed the max_length parameter and individual_duration parameter, i got the following result.

As you can see, the reconstructed music is still of length 33 second. And the final interpolated song has a playing sequence of 33 seconds and there is silence for 7 seconds and start playing again. Is there any other way to make max length of the constructed music dynamic so that there is no empty notes in between.

Hey everyone,
I have been trying to change this project to take multiple input (More than 2 midi files). Is there anyone who can provide me information regarding this section of this project?

song_one_trio =  1
song_two_trio =  0
start_trio = all_trios[0][song_one_trio]
end_trio = all_trios[1][song_two_trio]

trio_16bar_mean = interpolate(trio_models[trio_interp_model], start_trio, end_trio, num_steps=3, max_length=256, individual_duration=32, temperature=temperature)