File size: 4,096 Bytes
b607d8c
 
 
 
 
771bba0
b607d8c
 
 
 
771bba0
 
 
 
 
 
 
 
 
 
 
 
f524cd4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
771bba0
105058f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b607d8c
 
771bba0
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
<!doctype html>
<html>
	<head>
		<meta charset="utf-8" />
		<meta name="viewport" content="width=device-width" />
		<title>Melodiff MusicLDM v2</title>
		<link rel="stylesheet" href="style.css" />
	</head>
	<body>
		<div class="card">
			<h1>Melodiff MusicLDM v2</h1>
            <p>This is next version after <a href="https://huggingface.co/spaces/JanBabela/Riffusion-Melodiff-v1" target="_blank">Melodiff Riffusion v1</a> </p>
            <p>Melodiff MusicLDM continues to explore the idea of using the audio to audio pipeline of Stable Difussion audio models for creating cover versions of songs.</p>
			<p><br>Melodiff MusicLDM uses <a href="https://huggingface.co/ucsd-reach/musicldm" target="_blank">MusicLDM model</a> as base model for audio generation.</p>
            <p>What was done and what is presented here: Deconstructing the base pipeline and reconstructing back for audio to audio modifications.</p>
            <p>No new model training, nor finetuning was done, only modifications to base pipeline.</p>
            <p><br>MusicLDM generates audio of better quality compared to Riffusion (first) model. It generates samples of length 10s compared to 5s samples of previous model.</p>
            <p>Also speed of generation improved: previously it took about 8s to generate 5s long sample of mono audio. Now it takes about 8s to generate 10s long sample of  stereo audio.</p>
            <p>Also consistency. Previosly only about 30% of modified samples were good (or ok) and some prompt and seed play was needed to find good sound quality.</p>
            <p>Now about 70% of modified samples are good (or ok).</p>
            <p>Again longer modifications are possible by splitting, modifying and concatenating back the samples.</p>
            <p>Underlying MusicLDM model is two years old. It would be interesting to try new models, which have notably better quality.</p>          
            <p><br> Examples of music generated by modifying the underlying song: <br></p>
            <p>
            Bella Ciao, originally played by saxophone, modified to be played by electric guitar
            <audio controls>
              <source src="BellaElGuitar.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>
            <p>
            Bella Ciao, originally played by violin, modified to be played by piano
            <audio controls>
              <source src="BellaPiano.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>
            <p>
            Iko iko, originally played by saxophone, modified to be played by violin
            <audio controls>
              <source src="IkoViolin.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>
            <p>
            When the Saints, originally played by saxophone, modified to be played by strings
            <audio controls>
              <source src="SaintsStrings.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>            
            <p><br> Examples of original with modified samples: <br></p>
            <p>
            Saxophone solo, original
            <audio controls>
              <source src="MindscapeResampled.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>
            <p>
            Modified to be played by violin
            <audio controls>
              <source src="MindScapeViolin.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>
            <p>
            Modified to be played by electric guitar
            <audio controls>
              <source src="MindScapeElguitar.wav" type="audio/wav">
                Your browser does not support the audio element.
            </audio>
            </p>          
		</div>
	</body>
</html>