app.py

import gradio as gr
import numpy as np
from audioldm import text_to_audio, build_model
from share_btn import community_icon_html, loading_icon_html, share_js

model_id="haoheliu/AudioLDM-S-Full"

audioldm = None
current_model_name = None

# def predict(input, history=[]):
#     # tokenize the new input sentence
#     new_user_input_ids = tokenizer.encode(input + tokenizer.eos_token, return_tensors='pt')

#     # append the new user input tokens to the chat history
#     bot_input_ids = torch.cat([torch.LongTensor(history), new_user_input_ids], dim=-1)

#     # generate a response 
#     history = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id).tolist()

#     # convert the tokens to text, and then split the responses into lines
#     response = tokenizer.decode(history[0]).split("<|endoftext|>")
#     response = [(response[i], response[i+1]) for i in range(0, len(response)-1, 2)]  # convert to tuples of list
#     return response, history
  
def text2audio(text, duration, guidance_scale, random_seed, n_candidates, model_name="audioldm-m-text-ft"):
    global audioldm, current_model_name
    
    if audioldm is None or model_name != current_model_name:
        audioldm=build_model(model_name=model_name)
        current_model_name = model_name
        
    # print(text, length, guidance_scale)
    waveform = text_to_audio(
        latent_diffusion=audioldm,
        text=text,
        seed=random_seed,
        duration=duration,
        guidance_scale=guidance_scale,
        n_candidate_gen_per_text=int(n_candidates),
    )  # [bs, 1, samples]
    waveform = [
        gr.make_waveform((16000, wave[0]), bg_image="bg.png") for wave in waveform
    ]
    # waveform = [(16000, np.random.randn(16000)), (16000, np.random.randn(16000))]
    if(len(waveform) == 1):
      waveform = waveform[0]
    return waveform

iface = gr.Interface(fn=text2audio, inputs=[
        gr.Textbox(value="A man is speaking in a huge room", max_lines=1),
        gr.Slider(2.5, 10, value=5, step=2.5),
        gr.Slider(0, 5, value=2.5, step=0.5),
        gr.Number(value=42),
        gr.Number(value=3)
    ], outputs=[gr.Audio(label="Output", type="numpy"), gr.Audio(label="Output", type="numpy")],
                allow_flagging="never"
                     )
iface.launch(share=False)

#iface.queue(max_size=10).launch(debug=True)
# iface.launch(debug=True, share=True)