Update app.py
Browse files
app.py
CHANGED
|
@@ -4,33 +4,8 @@ from prompts import SYSTEM_PROMPT
|
|
| 4 |
from pydub import AudioSegment
|
| 5 |
import pypdf
|
| 6 |
import os
|
| 7 |
-
import io
|
| 8 |
import tempfile
|
| 9 |
|
| 10 |
-
def estimate_audio_length(text):
|
| 11 |
-
# Estimate 150 words per minute
|
| 12 |
-
word_count = len(text.split())
|
| 13 |
-
return word_count / 150 # Returns length in minutes
|
| 14 |
-
|
| 15 |
-
def trim_dialogue(dialogue, target_length_minutes):
|
| 16 |
-
trimmed_dialogue = []
|
| 17 |
-
current_length = 0
|
| 18 |
-
target_length_seconds = target_length_minutes * 60
|
| 19 |
-
|
| 20 |
-
for item in dialogue.dialogue:
|
| 21 |
-
item_length = estimate_audio_length(item.text)
|
| 22 |
-
if current_length + item_length > target_length_minutes:
|
| 23 |
-
# Trim this item to fit
|
| 24 |
-
words = item.text.split()
|
| 25 |
-
words_to_keep = int((target_length_minutes - current_length) * 150)
|
| 26 |
-
item.text = " ".join(words[:words_to_keep]) + "..."
|
| 27 |
-
trimmed_dialogue.append(item)
|
| 28 |
-
break
|
| 29 |
-
trimmed_dialogue.append(item)
|
| 30 |
-
current_length += item_length
|
| 31 |
-
|
| 32 |
-
return trimmed_dialogue
|
| 33 |
-
|
| 34 |
def generate_podcast(file, tone, length):
|
| 35 |
# Extract text from PDF
|
| 36 |
if not file.name.lower().endswith('.pdf'):
|
|
@@ -51,21 +26,15 @@ def generate_podcast(file, tone, length):
|
|
| 51 |
|
| 52 |
# Generate script
|
| 53 |
try:
|
| 54 |
-
script = generate_script(SYSTEM_PROMPT, truncated_text, tone)
|
| 55 |
except Exception as e:
|
| 56 |
raise gr.Error(f"Error generating script: {str(e)}")
|
| 57 |
|
| 58 |
-
# Determine target length in minutes
|
| 59 |
-
target_length = 2 if length == "Short (1-2 min)" else 5
|
| 60 |
-
|
| 61 |
-
# Trim dialogue to fit target length
|
| 62 |
-
trimmed_dialogue = trim_dialogue(script, target_length)
|
| 63 |
-
|
| 64 |
# Generate audio for each dialogue item
|
| 65 |
audio_segments = []
|
| 66 |
transcript = ""
|
| 67 |
try:
|
| 68 |
-
for item in
|
| 69 |
audio_file = generate_audio(item.text, item.speaker)
|
| 70 |
audio_segment = AudioSegment.from_mp3(audio_file)
|
| 71 |
audio_segments.append(audio_segment)
|
|
@@ -77,11 +46,6 @@ def generate_podcast(file, tone, length):
|
|
| 77 |
# Combine audio segments
|
| 78 |
combined_audio = sum(audio_segments)
|
| 79 |
|
| 80 |
-
# Ensure audio doesn't exceed target length
|
| 81 |
-
target_length_ms = target_length * 60 * 1000
|
| 82 |
-
if len(combined_audio) > target_length_ms:
|
| 83 |
-
combined_audio = combined_audio[:target_length_ms]
|
| 84 |
-
|
| 85 |
# Save combined audio to a temporary file
|
| 86 |
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as temp_audio:
|
| 87 |
combined_audio.export(temp_audio.name, format="mp3")
|
|
@@ -89,7 +53,7 @@ def generate_podcast(file, tone, length):
|
|
| 89 |
|
| 90 |
return temp_audio_path, transcript
|
| 91 |
|
| 92 |
-
# Gradio interface setup
|
| 93 |
instructions = """
|
| 94 |
# Podcast Generator
|
| 95 |
|
|
|
|
| 4 |
from pydub import AudioSegment
|
| 5 |
import pypdf
|
| 6 |
import os
|
|
|
|
| 7 |
import tempfile
|
| 8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
def generate_podcast(file, tone, length):
|
| 10 |
# Extract text from PDF
|
| 11 |
if not file.name.lower().endswith('.pdf'):
|
|
|
|
| 26 |
|
| 27 |
# Generate script
|
| 28 |
try:
|
| 29 |
+
script = generate_script(SYSTEM_PROMPT, truncated_text, tone, length)
|
| 30 |
except Exception as e:
|
| 31 |
raise gr.Error(f"Error generating script: {str(e)}")
|
| 32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
# Generate audio for each dialogue item
|
| 34 |
audio_segments = []
|
| 35 |
transcript = ""
|
| 36 |
try:
|
| 37 |
+
for item in script.dialogue:
|
| 38 |
audio_file = generate_audio(item.text, item.speaker)
|
| 39 |
audio_segment = AudioSegment.from_mp3(audio_file)
|
| 40 |
audio_segments.append(audio_segment)
|
|
|
|
| 46 |
# Combine audio segments
|
| 47 |
combined_audio = sum(audio_segments)
|
| 48 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
# Save combined audio to a temporary file
|
| 50 |
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as temp_audio:
|
| 51 |
combined_audio.export(temp_audio.name, format="mp3")
|
|
|
|
| 53 |
|
| 54 |
return temp_audio_path, transcript
|
| 55 |
|
| 56 |
+
# Gradio interface setup
|
| 57 |
instructions = """
|
| 58 |
# Podcast Generator
|
| 59 |
|