File size: 11,169 Bytes
ffead1e
 
 
ded8ad7
dec91cc
94c1252
559b00c
ffead1e
 
 
 
 
6134818
559b00c
dec91cc
 
559b00c
 
 
 
 
 
 
ffead1e
 
559b00c
ffead1e
559b00c
ffead1e
 
 
 
 
 
 
 
 
559b00c
 
 
ffead1e
 
 
 
 
 
 
 
 
 
 
559b00c
ffead1e
 
e206c8a
ffead1e
 
 
559b00c
e206c8a
ffead1e
559b00c
ffead1e
 
ded8ad7
ffead1e
559b00c
e206c8a
ffead1e
 
559b00c
ffead1e
559b00c
ffead1e
 
 
 
 
559b00c
ffead1e
 
a258601
559b00c
 
 
 
fa7808c
dec91cc
 
 
 
 
e206c8a
 
ded8ad7
e206c8a
dec91cc
 
 
e206c8a
 
 
ded8ad7
 
 
6134818
 
 
 
dec91cc
 
6134818
e206c8a
94c1252
e206c8a
 
ded8ad7
e206c8a
dec91cc
 
 
e206c8a
ded8ad7
6134818
dec91cc
 
6134818
dec91cc
 
6134818
dec91cc
6134818
dec91cc
 
 
ded8ad7
 
 
 
 
 
 
 
dec91cc
 
 
 
ded8ad7
ffead1e
05118bf
 
 
559b00c
 
1470851
559b00c
 
 
05118bf
559b00c
 
6de5b21
05118bf
559b00c
 
94c1252
 
dec91cc
559b00c
 
 
05118bf
 
e206c8a
dec91cc
bcb3137
dec91cc
e206c8a
dec91cc
 
ffead1e
6134818
fa7808c
dec91cc
 
 
ded8ad7
ffead1e
dec91cc
 
 
 
 
 
e206c8a
ded8ad7
e206c8a
dec91cc
 
 
6134818
 
ded8ad7
 
6134818
ded8ad7
dec91cc
 
ded8ad7
05118bf
ded8ad7
05118bf
dec91cc
 
 
05118bf
ded8ad7
 
 
 
05118bf
 
 
e206c8a
05118bf
 
ded8ad7
05118bf
dec91cc
 
 
05118bf
 
ded8ad7
 
 
 
05118bf
 
dec91cc
 
 
 
 
05118bf
7b7c424
05118bf
ded8ad7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
05118bf
 
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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
import gradio as gr
import json
import torch
import time
import random
import spaces

from tqdm import tqdm
from huggingface_hub import snapshot_download
from models import AudioDiffusion, DDPMScheduler
from audioldm.audio.stft import TacotronSTFT
from audioldm.variational_autoencoder import AutoencoderKL
from pydub import AudioSegment

max_64_bit_int = 2**63 - 1

# Automatic device detection
if torch.cuda.is_available():
    device_type = "cuda"
    device_selection = "cuda:0"
else:
    device_type = "cpu"
    device_selection = "cpu"

class Tango:
    def __init__(self, name = "declare-lab/tango2", device = device_selection):
        
        path = snapshot_download(repo_id = name)
        
        vae_config = json.load(open("{}/vae_config.json".format(path)))
        stft_config = json.load(open("{}/stft_config.json".format(path)))
        main_config = json.load(open("{}/main_config.json".format(path)))
        
        self.vae = AutoencoderKL(**vae_config).to(device)
        self.stft = TacotronSTFT(**stft_config).to(device)
        self.model = AudioDiffusion(**main_config).to(device)
        
        vae_weights = torch.load("{}/pytorch_model_vae.bin".format(path), map_location = device)
        stft_weights = torch.load("{}/pytorch_model_stft.bin".format(path), map_location = device)
        main_weights = torch.load("{}/pytorch_model_main.bin".format(path), map_location = device)
        
        self.vae.load_state_dict(vae_weights)
        self.stft.load_state_dict(stft_weights)
        self.model.load_state_dict(main_weights)

        print ("Successfully loaded checkpoint from:", name)
        
        self.vae.eval()
        self.stft.eval()
        self.model.eval()
        
        self.scheduler = DDPMScheduler.from_pretrained(main_config["scheduler_name"], subfolder = "scheduler")
        
    def chunks(self, lst, n):
        # Yield successive n-sized chunks from a list
        for i in range(0, len(lst), n):
            yield lst[i:i + n]
        
    def generate(self, prompt, steps = 100, guidance = 3, samples = 1, disable_progress = True):
        # Generate audio for a single prompt string
        with torch.no_grad():
            latents = self.model.inference([prompt], self.scheduler, steps, guidance, samples, disable_progress = disable_progress)
            mel = self.vae.decode_first_stage(latents)
            wave = self.vae.decode_to_waveform(mel)
        return wave
    
    def generate_for_batch(self, prompts, steps = 200, guidance = 3, samples = 1, batch_size = 8, disable_progress = True):
        # Generate audio for a list of prompt strings
        outputs = []
        for k in tqdm(range(0, len(prompts), batch_size)):
            batch = prompts[k: k + batch_size]
            with torch.no_grad():
                latents = self.model.inference(batch, self.scheduler, steps, guidance, samples, disable_progress = disable_progress)
                mel = self.vae.decode_first_stage(latents)
                wave = self.vae.decode_to_waveform(mel)
                outputs += [item for item in wave]
        if samples == 1:
            return outputs
        return list(self.chunks(outputs, samples))

# Initialize TANGO

tango = Tango(device = "cpu")
tango.vae.to(device_type)
tango.stft.to(device_type)
tango.model.to(device_type)

def update_seed(is_randomize_seed, seed):
    if is_randomize_seed:
        return random.randint(0, max_64_bit_int)
    return seed

def check(
    prompt,
    output_number,
    steps,
    guidance,
    is_randomize_seed,
    seed
):
    if prompt is None or prompt == "":
        raise gr.Error("Please provide a prompt input.")
    if not output_number in [1, 2, 3]:
        raise gr.Error("Please ask for 1, 2 or 3 output files.")

def update_output(output_format, output_number):
    return [
        gr.update(format = output_format),
        gr.update(format = output_format, visible = (2 <= output_number)),
        gr.update(format = output_format, visible = (output_number == 3)),
        gr.update(visible = False)
    ]

@spaces.GPU(duration=420)
def text2audio(
    prompt,
    output_number,
    steps,
    guidance,
    is_randomize_seed,
    seed
):
    start = time.time()

    if seed is None:
        seed = random.randint(0, max_64_bit_int)

    random.seed(seed)
    torch.manual_seed(seed)

    output_wave = tango.generate(prompt, steps, guidance, output_number)

    output_wave_1 = gr.make_waveform((16000, output_wave[0]))
    output_wave_2 = gr.make_waveform((16000, output_wave[1])) if (2 <= output_number) else None
    output_wave_3 = gr.make_waveform((16000, output_wave[2])) if (output_number == 3) else None

    end = time.time()
    secondes = int(end - start)
    minutes = secondes // 60
    secondes = secondes - (minutes * 60)
    hours = minutes // 60
    minutes = minutes - (hours * 60)
    return [
        output_wave_1,
        output_wave_2,
        output_wave_3,
        gr.update(visible = True, value = "Start again to get a different result. The output have been generated in " + ((str(hours) + " h, ") if hours != 0 else "") + ((str(minutes) + " min, ") if hours != 0 or minutes != 0 else "") + str(secondes) + " sec.")
    ]

# Gradio interface
with gr.Blocks() as interface:
    gr.Markdown("""
        <p style="text-align: center;">
        <b><big><big><big>Text-to-Audio</big></big></big></b>
        <br/>Generates 10 seconds of sound effects from description, freely, without account, without watermark.
        </p>
        <br/>
        <br/>
        ✨ Powered by <i>Tango 2</i> AI.
        <br/>
        <ul>
        <li>If you need <b>47 seconds</b> of audio, I recommend to use <i>Stable Audio</i>,</li>
        <li>If you need to generate <b>music</b>, I recommend to use <i>MusicGen</i>,</li>
        </ul>
        <br/>
        """ + ("πŸƒβ€β™€οΈ Estimated time: few minutes." if torch.cuda.is_available() else "🐌 Slow process... ~5 min.") + """
        Your computer must <b><u>not</u></b> enter into standby mode.<br/>You can duplicate this space on a free account, it's designed to work on CPU, GPU and ZeroGPU.<br/>
        <a href='https://huggingface.co/spaces/Fabrice-TIERCELIN/Text-to-Audio?duplicate=true&hidden=public&hidden=public'><img src='https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14'></a>
        <br/>
        βš–οΈ You can use, modify and share the generated sounds but not for commercial uses.
        """
    )
    input_text = gr.Textbox(label = "Prompt", value = "Snort of a horse", lines = 2, autofocus = True)
    with gr.Accordion("Advanced options", open = False):
        output_format = gr.Radio(label = "Output format", info = "The file you can dowload", choices = ["mp3", "wav"], value = "wav")
        output_number = gr.Slider(label = "Number of generations", info = "1, 2 or 3 output files", minimum = 1, maximum = 3, value = 1, step = 1, interactive = True)
        denoising_steps = gr.Slider(label = "Steps", info = "lower=faster & variant, higher=audio quality & similar", minimum = 10, maximum = 200, value = 10, step = 1, interactive = True)
        guidance_scale = gr.Slider(label = "Guidance Scale", info = "lower=audio quality, higher=follow the prompt", minimum = 1, maximum = 10, value = 3, step = 0.1, interactive = True)
        randomize_seed = gr.Checkbox(label = "\U0001F3B2 Randomize seed", value = True, info = "If checked, result is always different")
        seed = gr.Slider(minimum = 0, maximum = max_64_bit_int, step = 1, randomize = True, label = "Seed")

    submit = gr.Button("πŸš€ Generate", variant = "primary")

    output_audio_1 = gr.Audio(label = "Generated Audio #1/3", format = "wav", type="numpy", autoplay = True)
    output_audio_2 = gr.Audio(label = "Generated Audio #2/3", format = "wav", type="numpy")
    output_audio_3 = gr.Audio(label = "Generated Audio #3/3", format = "wav", type="numpy")
    information = gr.Label(label = "Information")

    submit.click(fn = update_seed, inputs = [
        randomize_seed,
        seed
    ], outputs = [
        seed
    ], queue = False, show_progress = False).then(fn = check, inputs = [
        input_text,
        output_number,
        denoising_steps,
        guidance_scale,
        randomize_seed,
        seed
    ], outputs = [], queue = False, show_progress = False).success(fn = update_output, inputs = [
        output_format,
        output_number
    ], outputs = [
        output_audio_1,
        output_audio_2,
        output_audio_3,
        information
    ], queue = False, show_progress = False).success(fn = text2audio, inputs = [
        input_text,
        output_number,
        denoising_steps,
        guidance_scale,
        randomize_seed,
        seed
    ], outputs = [
        output_audio_1,
        output_audio_2,
        output_audio_3,
        information
    ], scroll_to_output = True)

    gr.Examples(
        fn = text2audio,
	    inputs = [
            input_text,
            output_number,
            denoising_steps,
            guidance_scale,
            randomize_seed,
            seed
        ],
	    outputs = [
            output_audio_1,
            output_audio_2,
            output_audio_3,
            information
        ],
        examples = [
                ["A hammer is hitting a wooden surface", 3, 100, 3, False, 123],
                ["Peaceful and calming ambient music with singing bowl and other instruments.", 3, 100, 3, False, 123],
                ["A man is speaking in a small room.", 2, 100, 3, False, 123],
                ["A female is speaking followed by footstep sound", 1, 100, 3, False, 123],
                ["Wooden table tapping sound followed by water pouring sound.", 3, 200, 3, False, 123],
            ],
        cache_examples = "lazy",
    )
    
    gr.Markdown(
        """
        ## How to prompt your sound
        You can use round brackets to increase the importance of a part:
        ```
        Peaceful and (calming) ambient music with singing bowl and other instruments
        ```
        You can use several levels of round brackets to even more increase the importance of a part:
        ```
        (Peaceful) and ((calming)) ambient music with singing bowl and other instruments
        ```
        You can use number instead of several round brackets:
        ```
        (Peaceful:1.5) and ((calming)) ambient music with singing bowl and other instruments
        ```
        You can do the same thing with square brackets to decrease the importance of a part:
        ```
        (Peaceful:1.5) and ((calming)) ambient music with [singing:2] bowl and other instruments
        """
    )
        
    interface.queue(10).launch()