SBHacksXI/plot_music_sync.py at main · Alexander-Reyes9/SBHacksXI · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
import numpy as np
import librosa
import soundfile as sf
import matplotlib.pyplot as plt


def align_audio_with_dtw(x, wp, fs, hop_length=512):
    """
    Align an audio signal using a DTW warping path.

    Parameters:
    -----------
    x : np.ndarray
        Audio signal to align
    wp : np.ndarray
        Warping path (frames)
    fs : int
        Sampling rate
    hop_length : int, optional
        Hop length used for feature extraction

    Returns:
    --------
    np.ndarray
        Aligned audio signal
    """
    # Map frames to samples
    time_warped = librosa.frames_to_samples(wp[:, 1], hop_length=hop_length)
    time_original = librosa.frames_to_samples(wp[:, 0], hop_length=hop_length)

    # Interpolate the audio signal based on the warping path
    aligned_audio = np.interp(
        np.arange(time_original[-1] + 1),
        time_warped,
        x[:min(len(x), len(time_warped))],
    )
    return aligned_audio


def create_full_transition(x_1, x_2, wp, fs, crossfade_duration=1.0):
    """
    Create a full audio file containing the entire first track that transitions into the second track.

    Parameters:
    -----------
    x_1 : np.ndarray
        First audio signal (played in full)
    x_2 : np.ndarray
        Second audio signal (transitioned into)
    wp : np.ndarray
        Warping path (frames)
    fs : int
        Sampling rate
    crossfade_duration : float, optional
        Duration of the crossfade between tracks in seconds

    Returns:
    --------
    np.ndarray
        The complete audio signal with transition
    """
    # Convert crossfade duration to samples
    crossfade_samples = int(crossfade_duration * fs)

    # Align the second track using the warping path
    aligned_x2 = align_audio_with_dtw(x_2, wp, fs)

    # Ensure the first track is long enough for the crossfade
    transition_start = len(x_1) - crossfade_samples
    if transition_start < 0:
        raise ValueError("First audio signal is too short for the specified crossfade duration.")

    # Calculate the total length of the output signal
    total_length = max(len(x_1), transition_start + len(aligned_x2))

    # Create the output array
    output = np.zeros(total_length)

    # Copy the first signal
    output[:len(x_1)] = x_1

    # Create crossfade weights
    fade_out = np.linspace(1, 0, crossfade_samples)
    fade_in = np.linspace(0, 1, crossfade_samples)

    # Adjust lengths to match the transition region
    aligned_x2 = aligned_x2[:len(output) - transition_start]

    # Debugging shape mismatches
    print(f"Output shape: {output[transition_start:transition_start + crossfade_samples].shape}")
    print(f"Fade out shape: {fade_out.shape}")
    print(f"Aligned x2 fade in shape: {aligned_x2[:crossfade_samples].shape}")

    # Apply the crossfade
    output[transition_start:transition_start + crossfade_samples] *= fade_out
    aligned_x2[:crossfade_samples] *= fade_in

    # Add the second signal starting at the transition point
    output[transition_start:transition_start + len(aligned_x2)] += aligned_x2

    return output


def main():
    try:
        # Load audio files
        print("Loading audio files...")
        x_1, fs = librosa.load("bunny.mp3", sr=None)
        x_2, fs = librosa.load("music.mp3", sr=None)

        # Ensure mono audio
        if x_1.ndim > 1:
            x_1 = np.mean(x_1, axis=1)
        if x_2.ndim > 1:
            x_2 = np.mean(x_2, axis=1)

        print("Calculating chroma features...")
        # Calculate chroma features
        hop_length = 1024
        x_1_chroma = librosa.feature.chroma_cqt(y=x_1, sr=fs, hop_length=hop_length)
        x_2_chroma = librosa.feature.chroma_cqt(y=x_2, sr=fs, hop_length=hop_length)

        print("Computing DTW...")
        # Compute DTW
        D, wp = librosa.sequence.dtw(X=x_1_chroma, Y=x_2_chroma, metric="cosine")

        print("Creating transition...")
        # Create full audio with transition
        full_audio = create_full_transition(x_1, x_2, wp, fs, crossfade_duration=2.0)

        print("Saving output file...")
        # Save the result
        sf.write("full_transition.wav", full_audio, fs)
        print("Done! Output saved as 'full_transition.wav'")

    except Exception as e:
        print(f"An error occurred: {str(e)}")


if __name__ == "__main__":
    main()