Spaces:
Runtime error
Runtime error
vitorcalvi
commited on
Commit
•
d7529f8
1
Parent(s):
37b8131
pre-launch
Browse files- app.py +137 -0
- requirements.txt +9 -0
app.py
ADDED
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import PyPDF2
|
3 |
+
import nltk
|
4 |
+
from nltk.tokenize import sent_tokenize
|
5 |
+
from sumy.parsers.plaintext import PlaintextParser
|
6 |
+
from sumy.nlp.tokenizers import Tokenizer
|
7 |
+
from sumy.summarizers.lsa import LsaSummarizer
|
8 |
+
import os
|
9 |
+
from pydub import AudioSegment
|
10 |
+
from concurrent.futures import ThreadPoolExecutor
|
11 |
+
from TTS.api import TTS
|
12 |
+
|
13 |
+
# Download necessary NLTK data
|
14 |
+
nltk.download('punkt', quiet=True)
|
15 |
+
|
16 |
+
# Initialize TTS model using ONNX
|
17 |
+
tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", use_onnx=True)
|
18 |
+
|
19 |
+
# Set default speaker and language manually based on valid IDs obtained
|
20 |
+
default_speaker = "en_speaker_1" # Replace with a valid speaker ID from the printed list
|
21 |
+
default_language = "en" # Replace with a valid language code from the printed list
|
22 |
+
|
23 |
+
def extract_text_from_pdf(pdf_path):
|
24 |
+
try:
|
25 |
+
with open(pdf_path, 'rb') as file:
|
26 |
+
reader = PyPDF2.PdfReader(file)
|
27 |
+
text = ''
|
28 |
+
for page in reader.pages:
|
29 |
+
text += page.extract_text()
|
30 |
+
return text
|
31 |
+
except Exception as e:
|
32 |
+
print(f"Error extracting text from PDF: {e}")
|
33 |
+
return None
|
34 |
+
|
35 |
+
def summarize_text(text, summary_length):
|
36 |
+
parser = PlaintextParser.from_string(text, Tokenizer("english"))
|
37 |
+
summarizer = LsaSummarizer()
|
38 |
+
summary = summarizer(parser.document, summary_length)
|
39 |
+
return ' '.join([str(sentence) for sentence in summary])
|
40 |
+
|
41 |
+
def split_into_chapters(text, num_chapters):
|
42 |
+
sentences = sent_tokenize(text)
|
43 |
+
if len(sentences) <= num_chapters:
|
44 |
+
return sentences
|
45 |
+
sentences_per_chapter = max(1, len(sentences) // num_chapters)
|
46 |
+
chapters = []
|
47 |
+
for i in range(0, len(sentences), sentences_per_chapter):
|
48 |
+
chapter = ' '.join(sentences[i:i+sentences_per_chapter])
|
49 |
+
chapters.append(chapter)
|
50 |
+
while len(chapters) > num_chapters:
|
51 |
+
chapters[-2] += ' ' + chapters[-1]
|
52 |
+
chapters.pop()
|
53 |
+
return chapters
|
54 |
+
|
55 |
+
def text_to_speech(text, output_path, speaker, language):
|
56 |
+
tts.tts_to_file(text=text, file_path=output_path, speaker=speaker, language=language)
|
57 |
+
return output_path
|
58 |
+
|
59 |
+
def adjust_audio_speed(input_path, output_path, target_duration):
|
60 |
+
audio = AudioSegment.from_mp3(input_path)
|
61 |
+
current_duration = len(audio)
|
62 |
+
|
63 |
+
if current_duration == 0:
|
64 |
+
print(f"Warning: Audio file {input_path} has zero duration. Skipping speed adjustment.")
|
65 |
+
return input_path
|
66 |
+
|
67 |
+
speed_factor = current_duration / target_duration
|
68 |
+
if speed_factor < 0.1:
|
69 |
+
speed_factor = 0.1
|
70 |
+
|
71 |
+
try:
|
72 |
+
adjusted_audio = audio.speedup(playback_speed=speed_factor)
|
73 |
+
adjusted_audio.export(output_path, format="mp3")
|
74 |
+
return output_path
|
75 |
+
except Exception as e:
|
76 |
+
print(f"Error adjusting audio speed: {e}")
|
77 |
+
return input_path
|
78 |
+
|
79 |
+
def process_chapter(chapter, i, speaker, language):
|
80 |
+
try:
|
81 |
+
if len(chapter.strip()) == 0:
|
82 |
+
print(f"Warning: Chapter {i+1} is empty. Skipping.")
|
83 |
+
return None
|
84 |
+
|
85 |
+
temp_path = f"temp_chapter_{i+1}.mp3"
|
86 |
+
output_path = f"chapter_{i+1}.mp3"
|
87 |
+
|
88 |
+
text_to_speech(chapter, temp_path, speaker, language)
|
89 |
+
|
90 |
+
# Adjust speed to fit into 3 minutes
|
91 |
+
adjust_audio_speed(temp_path, output_path, 3 * 60 * 1000)
|
92 |
+
|
93 |
+
os.remove(temp_path) # Clean up temporary file
|
94 |
+
return output_path
|
95 |
+
except Exception as e:
|
96 |
+
print(f"Error processing chapter {i+1}: {e}")
|
97 |
+
return None
|
98 |
+
|
99 |
+
def process_pdf(pdf_path, num_chapters, speaker, language):
|
100 |
+
full_text = extract_text_from_pdf(pdf_path)
|
101 |
+
if full_text is None or len(full_text.strip()) == 0:
|
102 |
+
print("Error: Extracted text is empty or None")
|
103 |
+
return []
|
104 |
+
|
105 |
+
# Clean text to remove unwanted characters
|
106 |
+
full_text = full_text.replace('\t', ' ')
|
107 |
+
|
108 |
+
summary_length = max(1, 15 * 150 // len(full_text.split()))
|
109 |
+
summary = summarize_text(full_text, summary_length)
|
110 |
+
|
111 |
+
chapters = split_into_chapters(summary, num_chapters)
|
112 |
+
|
113 |
+
with ThreadPoolExecutor() as executor:
|
114 |
+
chapter_audios = list(executor.map(lambda i: process_chapter(chapters[i], i, speaker, language), range(len(chapters))))
|
115 |
+
|
116 |
+
return [audio for audio in chapter_audios if audio is not None]
|
117 |
+
|
118 |
+
def gradio_interface(pdf_file, num_chapters):
|
119 |
+
if pdf_file is None:
|
120 |
+
return [None] * 10
|
121 |
+
|
122 |
+
chapter_audios = process_pdf(pdf_file.name, num_chapters, default_speaker, default_language)
|
123 |
+
return chapter_audios + [None] * (10 - len(chapter_audios))
|
124 |
+
|
125 |
+
iface = gr.Interface(
|
126 |
+
fn=gradio_interface,
|
127 |
+
inputs=[
|
128 |
+
gr.File(label="Upload PDF Book"),
|
129 |
+
gr.Slider(minimum=1, maximum=10, step=1, label="Number of Chapters", value=5)
|
130 |
+
],
|
131 |
+
outputs=[gr.Audio(label=f"Chapter {i+1}") for i in range(10)],
|
132 |
+
title="PDF Book to Audiobook Summary",
|
133 |
+
description="Upload a PDF book to get a 15-minute audiobook summary split into chapters."
|
134 |
+
)
|
135 |
+
|
136 |
+
if __name__ == "__main__":
|
137 |
+
iface.launch(share=True)
|
requirements.txt
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
nltk
|
2 |
+
numpy
|
3 |
+
onnxruntime
|
4 |
+
PyPDF2
|
5 |
+
pydub
|
6 |
+
sumy
|
7 |
+
torch
|
8 |
+
TTS
|
9 |
+
gradio
|