File size: 6,237 Bytes
24f6ec0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
import importlib

from inspect import isfunction

import os
import soundfile as sf

def seed_everything(seed):
    import random, os
    import numpy as np
    import torch
    
    random.seed(seed)
    os.environ['PYTHONHASHSEED'] = str(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    torch.cuda.manual_seed(seed)
    torch.backends.cudnn.deterministic = True
    torch.backends.cudnn.benchmark = True
    
def save_wave(waveform, savepath, name="outwav"):
    if type(name) is not list:
        name = [name] * waveform.shape[0]

    for i in range(waveform.shape[0]):
        path = os.path.join(
            savepath,
            "%s_%s.wav"
            % (
                os.path.basename(name[i])
                if (not ".wav" in name[i])
                else os.path.basename(name[i]).split(".")[0],
                i,
            ),
        )
        sf.write(path, waveform[i, 0], samplerate=16000)

def exists(x):
    return x is not None


def default(val, d):
    if exists(val):
        return val
    return d() if isfunction(d) else d


def count_params(model, verbose=False):
    total_params = sum(p.numel() for p in model.parameters())
    if verbose:
        print(f"{model.__class__.__name__} has {total_params * 1.e-6:.2f} M params.")
    return total_params


def get_obj_from_str(string, reload=False):
    module, cls = string.rsplit(".", 1)
    if reload:
        module_imp = importlib.import_module(module)
        importlib.reload(module_imp)
    return getattr(importlib.import_module(module, package=None), cls)


def instantiate_from_config(config):
    if not "target" in config:
        if config == "__is_first_stage__":
            return None
        elif config == "__is_unconditional__":
            return None
        raise KeyError("Expected key `target` to instantiate.")
    return get_obj_from_str(config["target"])(**config.get("params", dict()))

def default_audioldm_config(model_name="audioldm-s-full"):    
    basic_config = {
        "wave_file_save_path": "./output",
        "id": {
            "version": "v1",
            "name": "default",
            "root": "/mnt/fast/nobackup/users/hl01486/projects/general_audio_generation/AudioLDM-python/config/default/latent_diffusion.yaml",
        },
        "preprocessing": {
            "audio": {"sampling_rate": 16000, "max_wav_value": 32768},
            "stft": {"filter_length": 1024, "hop_length": 160, "win_length": 1024},
            "mel": {
                "n_mel_channels": 64,
                "mel_fmin": 0,
                "mel_fmax": 8000,
                "freqm": 0,
                "timem": 0,
                "blur": False,
                "mean": -4.63,
                "std": 2.74,
                "target_length": 1024,
            },
        },
        "model": {
            "device": "cuda",
            "target": "audioldm.pipline.LatentDiffusion",
            "params": {
                "base_learning_rate": 5e-06,
                "linear_start": 0.0015,
                "linear_end": 0.0195,
                "num_timesteps_cond": 1,
                "log_every_t": 200,
                "timesteps": 1000,
                "first_stage_key": "fbank",
                "cond_stage_key": "waveform",
                "latent_t_size": 256,
                "latent_f_size": 16,
                "channels": 8,
                "cond_stage_trainable": True,
                "conditioning_key": "film",
                "monitor": "val/loss_simple_ema",
                "scale_by_std": True,
                "unet_config": {
                    "target": "audioldm.latent_diffusion.openaimodel.UNetModel",
                    "params": {
                        "image_size": 64,
                        "extra_film_condition_dim": 512,
                        "extra_film_use_concat": True,
                        "in_channels": 8,
                        "out_channels": 8,
                        "model_channels": 128,
                        "attention_resolutions": [8, 4, 2],
                        "num_res_blocks": 2,
                        "channel_mult": [1, 2, 3, 5],
                        "num_head_channels": 32,
                        "use_spatial_transformer": True,
                    },
                },
                "first_stage_config": {
                    "base_learning_rate": 4.5e-05,
                    "target": "audioldm.variational_autoencoder.autoencoder.AutoencoderKL",
                    "params": {
                        "monitor": "val/rec_loss",
                        "image_key": "fbank",
                        "subband": 1,
                        "embed_dim": 8,
                        "time_shuffle": 1,
                        "ddconfig": {
                            "double_z": True,
                            "z_channels": 8,
                            "resolution": 256,
                            "downsample_time": False,
                            "in_channels": 1,
                            "out_ch": 1,
                            "ch": 128,
                            "ch_mult": [1, 2, 4],
                            "num_res_blocks": 2,
                            "attn_resolutions": [],
                            "dropout": 0.0,
                        },
                    },
                },
                "cond_stage_config": {
                    "target": "audioldm.clap.encoders.CLAPAudioEmbeddingClassifierFreev2",
                    "params": {
                        "key": "waveform",
                        "sampling_rate": 16000,
                        "embed_mode": "audio",
                        "unconditional_prob": 0.1,
                    },
                },
            },
        },
    }
    
    if("-l-" in model_name):
        basic_config["model"]["params"]["unet_config"]["params"]["model_channels"] = 256
        basic_config["model"]["params"]["unet_config"]["params"]["num_head_channels"] = 64
    elif("-m-" in model_name):
        basic_config["model"]["params"]["unet_config"]["params"]["model_channels"] = 192
        basic_config["model"]["params"]["cond_stage_config"]["params"]["amodel"] = "HTSAT-base" # This model use a larger HTAST
        
    return basic_config