Spaces:
Running
on
T4
Running
on
T4
File size: 3,232 Bytes
9c20b4e 8bf1635 2e2148b 8bf1635 74e9bb4 8bf1635 9c20b4e 5c2ba64 9c20b4e a9922ff 5e89640 5c2ba64 9c20b4e 672cb3f 9c20b4e 5e89640 9c20b4e 311d9aa 5e89640 a9922ff 5c2ba64 9c20b4e 07ea011 9c20b4e 5c2ba64 9c20b4e 5c2ba64 9c20b4e 5c2ba64 9c20b4e 331caac 5c2ba64 9c20b4e 8729c75 9c20b4e fb76d6c 9c20b4e fb76d6c 9c20b4e ffbd52c 9c20b4e 5c2ba64 9c20b4e 5c2ba64 9c20b4e 1a8723f 9c20b4e 1a8723f 9c20b4e 1a8723f 5c2ba64 74e9bb4 5c2ba64 9c20b4e 5c2ba64 672cb3f db8ccb7 2e2148b 5c2ba64 fb76d6c 5c2ba64 3cf1f43 5e89640 5c2ba64 5e89640 |
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 |
"""
main.py
"""
# Standard library imports
import glob
import os
import time
from pathlib import Path
from tempfile import NamedTemporaryFile
from typing import List, Literal, Tuple
# Third-party imports
import gradio as gr
from fastapi import FastAPI
from fastapi.staticfiles import StaticFiles
from loguru import logger
from pydantic import BaseModel
from pypdf import PdfReader
from pydub import AudioSegment
# Local imports
from prompts import SYSTEM_PROMPT
from utils import generate_script, generate_audio
app = FastAPI()
app.mount("/static", StaticFiles(directory="static"), name="static")
class DialogueItem(BaseModel):
"""A single dialogue item."""
speaker: Literal["Host (Jane)", "Guest"]
text: str
class Dialogue(BaseModel):
"""The dialogue between the host and guest."""
scratchpad: str
participants: List[str]
dialogue: List[DialogueItem]
def generate_podcast(file: str) -> Tuple[str, str]:
"""Generate the audio and transcript from the PDF."""
# Read the PDF file and extract text
with Path(file).open("rb") as f:
reader = PdfReader(f)
text = "\n\n".join([page.extract_text() for page in reader.pages])
# Call the LLM
llm_output = generate_script(SYSTEM_PROMPT, text, Dialogue)
logger.info(f"Generated dialogue: {llm_output}")
# Process the dialogue
audio_segments = []
transcript = ""
total_characters = 0
for line in llm_output.dialogue:
logger.info(f"Generating audio for {line.speaker}: {line.text}")
transcript_line = f"{line.speaker}: {line.text}"
transcript += transcript_line + "\n\n"
total_characters += len(line.text)
# Get audio file path
audio_file_path = generate_audio(line.text, line.speaker)
# Read the audio file into an AudioSegment
audio_segment = AudioSegment.from_file(audio_file_path)
audio_segments.append(audio_segment)
# Concatenate all audio segments
combined_audio = sum(audio_segments)
# Export the combined audio to a temporary file
temporary_directory = "./gradio_cached_examples/tmp/"
os.makedirs(temporary_directory, exist_ok=True)
temporary_file = NamedTemporaryFile(
dir=temporary_directory,
delete=False,
suffix=".mp3",
)
combined_audio.export(temporary_file.name, format="mp3")
# Delete any files in the temp directory that end with .mp3 and are over a day old
for file in glob.glob(f"{temporary_directory}*.mp3"):
if os.path.isfile(file) and time.time() - os.path.getmtime(file) > 24 * 60 * 60:
os.remove(file)
logger.info(f"Generated {total_characters} characters of audio")
return temporary_file.name, transcript
demo = gr.Interface(
title="OpenPodcast",
description="Convert your PDFs into podcasts with open-source AI models.",
fn=generate_podcast,
inputs=[
gr.File(
label="PDF",
),
],
outputs=[
gr.Audio(label="Audio", format="mp3"),
gr.Textbox(label="Transcript"),
],
allow_flagging="never",
api_name=False,
)
app = gr.mount_gradio_app(app, demo, path="/")
if __name__ == "__main__":
demo.launch(show_api=False)
|