Spaces:
Sleeping
Sleeping
File size: 1,780 Bytes
0c97faa 1e937ce db2ba20 1e937ce d2e67c6 1e937ce 0c97faa ccadd27 db2ba20 d67a1b2 d2e67c6 1e937ce d2e67c6 1e937ce d2e67c6 1e937ce d2e67c6 0c97faa |
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 |
import gradio as gr
import torch
from huggingface_hub import InferenceClient
from transformers import BarkModel
from transformers import AutoProcessor
model = BarkModel.from_pretrained("suno/bark-small")
device = "cuda:0" if torch.cuda.is_available() else "cpu"
model = model.to(device)
processor = AutoProcessor.from_pretrained("suno/bark")
"""
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")
with gr.Blocks() as demo:
chatbot = gr.Chatbot(type="messages")
audio_box = gr.Audio(autoplay=True)
msg = gr.Textbox(submit_btn=True)
clear = gr.Button("Clear")
def user(user_message, history: list):
return "", history + [{"role": "user", "content": user_message}]
def bot(history: list):
history.append({"role": "assistant", "content": ""})
for message in client.chat_completion(
history,
stream=True,
):
token = message.choices[0].delta.content
history[-1]["content"] += token
yield history
return history
def read(history: list):
text = history[-1]["content"]
inputs = processor(text=text, return_tensors="pt").to(device)
speech = model.generate(**inputs.to(device))
sampling_rate = model.generation_config.sample_rate
return tuple((sampling_rate, speech.cpu().numpy().squeeze()))
msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
bot, chatbot, chatbot
).then(read, chatbot, audio_box)
clear.click(lambda: None, None, chatbot, queue=False)
if __name__ == "__main__":
demo.launch()
|