Skroll / app.py
Afeezee's picture
Create app.py
ba9ada9 verified
import gradio as gr
import ffmpeg
import os
import docx
import warnings
import assemblyai as aai
import subprocess
# Suppress FutureWarnings
warnings.simplefilter("ignore", category=FutureWarning)
Key = os.getenv("KeyA") # Ensure this is set in your environment
aai.settings.api_key = Key
# Function to check if FFmpeg is installed
def is_ffmpeg_installed():
try:
subprocess.run(["ffmpeg", "-version"], stdout=subprocess.PIPE, stderr=subprocess.PIPE, check=True)
return True
except subprocess.CalledProcessError:
return False
except FileNotFoundError:
return False
# Function to extract audio from video safely
def extract_audio(video_path, output_audio_path="temp_audio.mp3"):
if not is_ffmpeg_installed():
raise RuntimeError("FFmpeg is not installed or not found in PATH.")
try:
ffmpeg.input(video_path).output(output_audio_path, format="mp3").run(overwrite_output=True, quiet=True)
return output_audio_path
except ffmpeg.Error as e:
raise RuntimeError(f"FFmpeg error: {e.stderr.decode()}")
# Function to transcribe audio using AssemblyAI
def transcribe_audio(file):
ext = os.path.splitext(file.name)[-1].lower()
audio_path = "temp_audio.mp3"
# Extract audio if video is uploaded
if ext in [".mp4", ".avi", ".mov", ".mkv"]:
audio_path = extract_audio(file.name)
else:
audio_path = file.name # Use audio file directly
# Upload file to AssemblyAI
transcriber = aai.Transcriber()
config = aai.TranscriptionConfig(speaker_labels=True)
transcript = transcriber.transcribe(audio_path, config=config)
return "\n".join([f"Speaker {utt.speaker}: {utt.text}" for utt in transcript.utterances])
# Function to export transcription
def save_transcription(text, file_format):
file_path = f"transcription.{file_format.lower()}"
if file_format == "TXT":
with open(file_path, "w") as f:
f.write(text)
elif file_format == "DOCX":
doc = docx.Document()
doc.add_paragraph(text)
doc.save(file_path)
elif file_format == "SRT":
with open(file_path, "w") as f:
for i, line in enumerate(text.split(".")):
start_time = f"00:00:{i*5:02d},000"
end_time = f"00:00:{(i+1)*5:02d},000"
f.write(f"{i+1}\n{start_time} --> {end_time}\n{line.strip()}\n\n")
return file_path
# Gradio Interface
with gr.Blocks() as demo:
gr.Markdown("# 🎙️ Skroll - Audio & Video Transcription Tool")
gr.Markdown("Upload an audio or video file and transcribe. Export in .txt, .docx, or .srt format.")
file_input = gr.File(label="Upload Audio or Video")
transcript_output = gr.Textbox(label="Transcription", interactive=True, lines=10)
transcribe_btn = gr.Button("Transcribe")
with gr.Row():
file_format = gr.Dropdown(["TXT", "DOCX", "SRT"], label="Export Format")
export_btn = gr.Button("Export")
download_link = gr.File(label="Download Transcription")
# Define Actions
transcribe_btn.click(transcribe_audio, inputs=[file_input], outputs=transcript_output)
export_btn.click(save_transcription, inputs=[transcript_output, file_format], outputs=download_link)
# Launch App
demo.launch(debug=True)