audio-cpp commited on
Commit
f80717a
·
verified ·
1 Parent(s): e36cfad

Update Yue2 README demos

Browse files
Files changed (1) hide show
  1. README.md +105 -3
README.md CHANGED
@@ -29,10 +29,112 @@ testing, validation, and optimization. Star our repo so you don’t miss any imp
29
  | `sidecars/yue2-generation-config.json` | Yue2 generation defaults. |
30
  | `sidecars/yue2-qwen.tiktoken` | Qwen tokenizer used by Yue2. |
31
  | `sidecars/yue2-vae-config.json` | Yue2 VAE config. |
32
- | `examples/tonight-awake-audiocpp-current.wav` | audio.cpp output from the official `tonight-awake` example. |
 
 
 
33
 
34
- The example WAV was generated using audio.cpp from the official Yue2 example:
35
- [`examples/tonight-awake.json`](https://huggingface.co/m-a-p/YuE2-3B/blob/main/examples/tonight-awake.json).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36
 
37
  ## Upstream
38
 
 
29
  | `sidecars/yue2-generation-config.json` | Yue2 generation defaults. |
30
  | `sidecars/yue2-qwen.tiktoken` | Qwen tokenizer used by Yue2. |
31
  | `sidecars/yue2-vae-config.json` | Yue2 VAE config. |
32
+ | `examples/demo1.wav` | Direct text-to-music demo, `cot=off`. |
33
+ | `examples/demo2.wav` | Full planning demo, `cot=full`. |
34
+ | `examples/demo3.wav` | Melody-conditioned cover demo, `cot=melody`. |
35
+ | `examples/demo4.wav` | Full score-conditioned demo, `cot=full`. |
36
 
37
+ ## Demos
38
+
39
+ The demos below were generated with `yue2-3b-bf16.gguf` and `yue2-vae-f32.gguf`.
40
+ Replace `<AUDIOCPP>` with your audio.cpp checkout and `<YUE2_GGUF>` with this
41
+ downloaded GGUF repo directory.
42
+
43
+ ### Demo 1: direct generation
44
+
45
+ <audio controls src="examples/demo1.wav"></audio>
46
+
47
+ ```bash
48
+ <AUDIOCPP>/build/debug/bin/audiocpp_cli \
49
+ --task gen \
50
+ --family yue2 \
51
+ --model <YUE2_GGUF> \
52
+ --backend cuda \
53
+ --threads 8 \
54
+ --text $'[Verse]\nSoft morning light is touching the window.\nI hear the city waking below.\n[Chorus]\nStay with the rhythm, let it carry us home.\nSing with the sunrise, we are never alone.' \
55
+ --request-option 'style=English, indie pop, bright acoustic guitar, soft drums, warm lead vocal, polished demo mix' \
56
+ --request-option cot=off \
57
+ --request-option seed=20260920 \
58
+ --request-option num_inference_steps=8 \
59
+ --session-option yue2.model_gguf=yue2-3b-bf16.gguf \
60
+ --session-option yue2.vae_gguf=yue2-vae-f32.gguf \
61
+ --out demo1.wav \
62
+ --log
63
+ ```
64
+
65
+ ### Demo 2: full planning
66
+
67
+ <audio controls src="examples/demo2.wav"></audio>
68
+
69
+ ```bash
70
+ <AUDIOCPP>/build/debug/bin/audiocpp_cli \
71
+ --task gen \
72
+ --family yue2 \
73
+ --model <YUE2_GGUF> \
74
+ --backend cuda \
75
+ --threads 8 \
76
+ --text $'[Verse]\nA quiet piano opens the evening.\nWarm city lights are starting to glow.\n[Chorus]\nLift every voice into the skyline.\nLet the melody carry us home.' \
77
+ --request-option 'style=English, piano pop, clear lead vocal, gentle bass, soft drums, warm chorus harmonies' \
78
+ --request-option cot=full \
79
+ --request-option seed=20260921 \
80
+ --request-option num_inference_steps=8 \
81
+ --session-option yue2.model_gguf=yue2-3b-bf16.gguf \
82
+ --session-option yue2.vae_gguf=yue2-vae-f32.gguf \
83
+ --out demo2.wav \
84
+ --log
85
+ ```
86
+
87
+ ### Demo 3: melody-conditioned cover
88
+
89
+ <audio controls src="examples/demo3.wav"></audio>
90
+
91
+ This demo uses the upstream
92
+ [`examples/melody.abc`](https://huggingface.co/m-a-p/YuE2-3B/blob/main/examples/melody.abc).
93
+
94
+ ```bash
95
+ <AUDIOCPP>/build/debug/bin/audiocpp_cli \
96
+ --task gen \
97
+ --family yue2 \
98
+ --model <YUE2_GGUF> \
99
+ --backend cuda \
100
+ --threads 8 \
101
+ --text $'[Verse]\nWe follow the melody line.\nThe rhythm keeps everything fine.\n[Chorus]\nHold that shape and make it new.\nChange the color, keep the view.' \
102
+ --request-option 'style=English, jazz funk cover, warm Rhodes, round bass, light drums, relaxed vocal, clean live band feel' \
103
+ --request-option cot=melody \
104
+ --request-option abc_file=<AUDIOCPP>/reference/YuE/examples/melody.abc \
105
+ --request-option seed=20260922 \
106
+ --request-option num_inference_steps=8 \
107
+ --session-option yue2.model_gguf=yue2-3b-bf16.gguf \
108
+ --session-option yue2.vae_gguf=yue2-vae-f32.gguf \
109
+ --out demo3.wav \
110
+ --log
111
+ ```
112
+
113
+ ### Demo 4: full score conditioning
114
+
115
+ <audio controls src="examples/demo4.wav"></audio>
116
+
117
+ This demo uses the upstream
118
+ [`examples/score-jazz.abc`](https://github.com/multimodal-art-projection/YuE/blob/main/examples/score-jazz.abc).
119
+
120
+ ```bash
121
+ <AUDIOCPP>/build/debug/bin/audiocpp_cli \
122
+ --task gen \
123
+ --family yue2 \
124
+ --model <YUE2_GGUF> \
125
+ --backend cuda \
126
+ --threads 8 \
127
+ --text $'[Verse]\nThe room begins with a gentle phrase.\nThe harmony turns through a golden haze.\n[Chorus]\nHold the note and let it rise.\nMorning opens up the sky.' \
128
+ --request-option 'style=English, piano pop with jazz harmony, clear lead vocal, gentle bass, tasteful drum fills' \
129
+ --request-option cot=full \
130
+ --request-option abc_file=<AUDIOCPP>/reference/YuE/examples/score-jazz.abc \
131
+ --request-option seed=20260923 \
132
+ --request-option num_inference_steps=8 \
133
+ --session-option yue2.model_gguf=yue2-3b-bf16.gguf \
134
+ --session-option yue2.vae_gguf=yue2-vae-f32.gguf \
135
+ --out demo4.wav \
136
+ --log
137
+ ```
138
 
139
  ## Upstream
140