import gradio as gr
from transformers import pipeline
import os
import torch
# Set max_split_size_mb
os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:50'
title = """# 🙋🏻♂️Welcome to🌟Tonic's Nexus🐦⬛Raven"""
description = """You can build with this endpoint using Nexus Raven. The demo is still a work in progress but we hope to add some endpoints for commonly used functions such as intention mappers and audiobook processing.
You can also use Nexus🐦⬛Raven on your laptop & by cloning this space. 🧬🔬🔍 Simply click here: 
Join us : 🌟TeamTonic🌟 is always making cool demos! Join our active builder's🛠️community on 👻Discord: [Discord](https://discord.gg/GWpVpekp) On 🤗Huggingface: [TeamTonic](https://huggingface.co./TeamTonic) & [MultiTransformer](https://huggingface.co./MultiTransformer) On 🌐Github: [Polytonic](https://github.com/tonic-ai) & contribute to 🌟 [PolyGPT](https://github.com/tonic-ai/polygpt-alpha)
"""
raven_pipeline = pipeline(
"text-generation",
model="Nexusflow/NexusRaven-V2-13B",
torch_dtype="auto",
device_map="auto",
)
class DialogueToSpeechConverter:
def __init__(self):
self.raven_pipeline = raven_pipeline
def process_text(self, input_text: str) -> str:
prompt = f"User Query: {input_text}"
result = self.raven_pipeline(prompt, max_new_tokens=2048, return_full_text=False, do_sample=False, temperature=0.001)[0]["generated_text"]
torch.cuda.empty_cache()
return result
def create_interface():
converter = DialogueToSpeechConverter()
with gr.Blocks() as app:
gr.Markdown(title)
gr.Markdown(description)
with gr.Row():
input_text = gr.Textbox(label="Input Text")
submit_button = gr.Button("Submit")
output_text = gr.Textbox(label="Nexus🐦⬛Raven")
submit_button.click(converter.process_text, inputs=input_text, outputs=output_text)
return app
if __name__ == "__main__":
converter = DialogueToSpeechConverter()
demo = gr.Interface(
fn=converter.process_text,
inputs="text",
outputs="text",
examples=[
['''
Function:
def create_audio_sequence_order(text):
"""
Analyzes the text and creates an order for each character and narrator segment.
Args:
text (str): The text containing the dialogues and narration.
Returns:
list: A list of tuples, each containing the character/narrator name and a segment of their dialogue/narration.
"""
Function:
def convert_text_to_speech_single_voice(text, voice):
"""
Converts a given text to speech using a specified voice. This function is used when there is only one character in the text.
Args:
text (str): The text to be converted to speech.
voice (str): The voice to be used for the audio generation.
Returns:
str: The path to the generated speech MP3 file.
"""
User Query: Currently, one way that the wealthy distinguish themselves from others is through the collection of rare objects. In a Celebration Society, to own an “original” of something will remain significant. However, barring a desire to prevent others from enjoying the experience, it will become possible to have perfect replicas of all manner of objects including paintings and sculptures.
There will still be pride of ownership in the original. Others will be able to fully enjoy the “same” piece as well.
use either speech to single voice if there's no dialogue or create_audio_sequence_order if there is dialogue
''']
],
title=title,
description=description
)
demo.launch()