Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import torch
|
| 3 |
+
from transformers import pipeline
|
| 4 |
+
|
| 5 |
+
MODEL = "openai/whisper-tiny" # good balance of speed + accuracy
|
| 6 |
+
|
| 7 |
+
# Load model once at startup
|
| 8 |
+
device = 0 if torch.cuda.is_available() else -1
|
| 9 |
+
|
| 10 |
+
transcriber = pipeline(
|
| 11 |
+
"automatic-speech-recognition",
|
| 12 |
+
model=MODEL,
|
| 13 |
+
device=device
|
| 14 |
+
)
|
| 15 |
+
|
| 16 |
+
def transcribe(audio):
|
| 17 |
+
if audio is None:
|
| 18 |
+
return "Please upload or record audio."
|
| 19 |
+
|
| 20 |
+
result = transcriber(audio)
|
| 21 |
+
return result["text"]
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
demo = gr.Interface(
|
| 25 |
+
fn=transcribe,
|
| 26 |
+
inputs=gr.Audio(
|
| 27 |
+
sources=["microphone", "upload"],
|
| 28 |
+
type="filepath"
|
| 29 |
+
),
|
| 30 |
+
outputs=gr.Textbox(label="Transcription"),
|
| 31 |
+
title="🎙 Voice to Text",
|
| 32 |
+
description="Upload or record audio. Powered by Whisper running locally in HF Spaces."
|
| 33 |
+
)
|
| 34 |
+
|
| 35 |
+
if __name__ == "__main__":
|
| 36 |
+
demo.launch()
|