|
1 | 1 | --- |
2 | 2 | layout: page |
3 | | -title: Listening Test Examples |
| 3 | +title: Audio Examples |
4 | 4 | --- |
5 | 5 | <head> |
6 | 6 | <style> |
@@ -40,75 +40,28 @@ title: Listening Test Examples |
40 | 40 |
|
41 | 41 |
|
42 | 42 | <div class="page"> |
43 | | - <h2>Listening Test - Realism Evaluation</h2> |
44 | | - <p>We conducted a <strong>realism evaluation</strong> to focusing on how well the pitch expression and overall timbre of the synthesized samples matched those of actual violin performances. </p> |
| 43 | + <h2>Vocoder Quality</h2> |
| 44 | + <p> Here, we provide examples to illustrate the quality of the Soundstream vocoder, which sets an upper bound on the quality of our generated performances. </p> |
45 | 45 |
|
46 | | - <p>The evaluation was carried out using the MUSHRA <cite>{1}</cite> protocol, a standardized tool for subjective audio quality assessment. Participants rated multiple models, including:</p> |
47 | | - |
48 | | - <ul> |
49 | | - <li><strong>ViolinDiff</strong>: Our Proposed model</li> |
50 | | - <li><strong>NoBend Model</strong>: Against a version of DiffSynth trained without pitch bend information</li> |
51 | | - <li><strong>Hawthorne et al.</strong> <cite>{2}</cite>: A multi-instrument diffusion-based model focused on synthesizing various instruments in polyphonic contexts.</li> |
52 | | - <li><strong>Maman et al.</strong> <cite>{3}</cite>: A model using performer embeddings to better capture timbre and style in orchestral instrument synthesis.</li> |
53 | | - <li><strong>GM Soundfont (Low Anchor)</strong>: The lowest quality baseline, serving as an anchor point for comparison.</li> |
54 | | - </ul> |
55 | | - |
56 | | - <h3>References</h3> |
57 | | - <ol> |
58 | | - <li>ITU, “Method for the subjective assessment of intermediate quality level of audio systems,” BS.1534, 2014.</li> |
59 | | - <li>Hawthorne et al.“Multi-instrument music synthesis with spectrogram diffusion,” in Proceedings of the ISMIR, 2022</li> |
60 | | - <li>Maman et al. "Performance conditioning for diffusion-based multi-instrument music synthesis,” in ICASSP, 2024</li> |
61 | | - </ol> |
62 | | - |
63 | | -<p>We would like to extend our gratitude to Ben Maman for providing the essential test files, which greatly contributed to this evaluation.</p> |
64 | 46 |
|
65 | 47 |
|
66 | 48 | <section class="audio-comparison"> |
67 | 49 | <h2>Kayser_Op20-36</h2> |
68 | 50 | <div class="audio-container"> |
69 | 51 | <div class="audio-block"> |
70 | | - <p>Original Audio</p> |
71 | | - <audio controls> |
72 | | - <source src="original_segment_Kayser_Op20-36_performer_2.wav"> |
73 | | - Your browser does not support the audio element. |
74 | | - </audio> |
75 | | - </div> |
76 | | - <div class="audio-block"> |
77 | | - <p>ViolinDiff</p> |
78 | | - <audio controls> |
79 | | - <source src="bend_segment_Kayser_Op20-36_performer_2.wav"> |
80 | | - Your browser does not support the audio element. |
81 | | - </audio> |
82 | | - </div> |
83 | | - <div class="audio-block"> |
84 | | - <p>NoBend Model</p> |
85 | | - <audio controls> |
86 | | - <source src="no_bend_segment_Kayser_Op20-36_performer_2.wav"> |
87 | | - Your browser does not support the audio element. |
88 | | - </audio> |
89 | | - </div> |
90 | | - <div class="audio-block"> |
91 | | - <p>Hawth.</p> |
| 52 | + <p>Original</p> |
92 | 53 | <audio controls> |
93 | | - <source src="magenta_segment_Kayser_Op20-36_performer_2.wav"> |
| 54 | + <source src="/vocoder/Kayser_Op20-36_org_7.wav"> |
94 | 55 | Your browser does not support the audio element. |
95 | 56 | </audio> |
96 | 57 | </div> |
97 | 58 | <div class="audio-block"> |
98 | | - <p>Maman.</p> |
| 59 | + <p>Vocoder</p> |
99 | 60 | <audio controls> |
100 | | - <source src="perfdiff_segment_Kayser_Op20-36_performer_2.wav"> |
| 61 | + <source src="/vocoder/Kayser_Op20-36_vocoder_7.wav"> |
101 | 62 | Your browser does not support the audio element. |
102 | 63 | </audio> |
103 | 64 | </div> |
104 | | - <div class="audio-block"> |
105 | | - <p>GM</p> |
106 | | - <audio controls> |
107 | | - <source src="window_gm_segment_Kayser_Op20-36_performer_2.wav"> |
108 | | - Your browser does not support the audio element. |
109 | | - </audio> |
110 | | - </div> |
111 | | - </div> |
112 | 65 | </section> |
113 | 66 |
|
114 | 67 | <section class="audio-comparison"> |
|
0 commit comments