Spaces:
Sleeping
Sleeping
File size: 2,750 Bytes
99e2ed9 |
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 |
import gradio as gr
from transformers import pipeline
import torch
from pydub import AudioSegment
import os
# Initialize the Whisper model
try:
whisper = pipeline(
"automatic-speech-recognition",
model="openai/whisper-small",
device="cuda" if torch.cuda.is_available() else "cpu"
)
except Exception as e:
raise Exception(f"Failed to load Whisper model: {str(e)}")
# Define the transcription function with chunking and automatic language detection
def transcribe_audio(audio):
if audio is None:
return "Error: Please upload an audio file."
# Validate file size (100 MB limit)
try:
file_size_mb = os.path.getsize(audio) / (1024 * 1024)
if file_size_mb > 100:
return "Error: Audio file exceeds 100 MB limit."
except FileNotFoundError:
return "Error: Audio file not found."
try:
# Load and process audio
audio_segment = AudioSegment.from_file(audio)
duration_ms = len(audio_segment)
chunk_length_ms = 30000 # 30 seconds
# Chunk long audio files
if duration_ms > chunk_length_ms:
chunks = [audio_segment[i:i + chunk_length_ms] for i in range(0, duration_ms, chunk_length_ms)]
transcriptions = []
for i, chunk in enumerate(chunks):
chunk_path = f"chunk_{i}.wav"
chunk.export(chunk_path, format="wav")
result = whisper(chunk_path, generate_kwargs={"task": "transcribe"}) # Automatic language detection
transcriptions.append(result["text"])
if os.path.exists(chunk_path):
os.remove(chunk_path)
return " ".join(transcriptions)
else:
result = whisper(audio, generate_kwargs={"task": "transcribe"}) # Automatic language detection
return result["text"]
except Exception as e:
return f"Error during transcription: {str(e)}"
finally:
# Clean up uploaded file
if os.path.exists(audio):
try:
os.remove(audio)
except Exception:
pass
# Create Gradio interface
demo = gr.Interface(
fn=transcribe_audio,
inputs=[
gr.Audio(type="filepath", label="Upload an Audio File (MP3, WAV, max 100 MB)")
],
outputs=gr.Textbox(label="Transcription"),
title="Audio to Text Transcription with Whisper",
description="Upload an audio file (MP3/WAV, up to 100 MB) to transcribe it using Open AI's Whisper model with automatic language detection.",
allow_flagging="never"
)
# Launch the app
if __name__ == "__main__":
demo.launch() |