import gradio import webui.modules.implementations.audioldm2 as aldm2 from webui.modules import util def generate(prompt, negative, duration, steps, cfg, seed, progress=gradio.Progress()): output = aldm2.generate(prompt, negative, steps, duration, cfg, seed, callback=lambda step, _, _2: progress((step, steps), desc='Generating...')) if isinstance(output, str): return None, None, output else: return output[1], util.make_waveform(output[1]), f'Successfully generated audio with seed: {output[0]}.' def audioldm2_tab(): with gradio.Row(): with gradio.Row(): selected = gradio.Dropdown(aldm2.models, value='cvssp/audioldm2', label='Model') with gradio.Column(elem_classes='smallsplit'): load_button = gradio.Button('🚀', variant='tool secondary') unload_button = gradio.Button('💣', variant='tool primary') def load(model): aldm2.create_model(model) return gradio.update() def unload(): aldm2.delete_model() return gradio.update() load_button.click(fn=load, inputs=selected, outputs=selected, show_progress=True) unload_button.click(fn=unload, outputs=selected, show_progress=True) with gradio.Row(): gen_button = gradio.Button('Generate', variant='primary') with gradio.Row(): with gradio.Column(): prompt = gradio.TextArea(label='Prompt', info='Put the audio you want here.', placeholder='The sound of a hammer hitting a wooden surface') neg_prompt = gradio.TextArea(label='Negative prompt', info='Put things to avoid generating here.', placeholder='low bitrate, low quality, bad quality') duration = gradio.Number(5, label='Duration (s)', info='Duration for the generation in seconds.') seed = gradio.Number(-1, label='Seed', info='Default: -1 (random). Set the seed for generation, random seed is used when a negative number is given.', precision=0) with gradio.Accordion('➕ Extra options', open=False): cfg = gradio.Slider(1, 20, 3.5, step=0.01, label='CFG scale', info='Default: 2.5. How much should the prompt affect the audio on every step?') steps = gradio.Slider(1, 300, 50, step=1, label='Steps', info='Default: 10. How many diffusion steps should be performed?') with gradio.Column(): with gradio.Row(): audio_out = gradio.Audio(label='Generated audio', interactive=False) with gradio.Row(): video_out = gradio.Video(label='Waveform video', interactive=False) with gradio.Row(): text_out = gradio.Textbox(label='Result') gen_button.click(fn=generate, inputs=[prompt, neg_prompt, duration, steps, cfg, seed], outputs=[audio_out, video_out, text_out])