Update app.py
Browse files
app.py
CHANGED
@@ -1,19 +1,33 @@
|
|
1 |
import gradio as gr
|
|
|
2 |
import whisper
|
3 |
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
|
9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
# Interface for Gradio
|
12 |
iface = gr.Interface(
|
13 |
-
fn=
|
14 |
inputs=audio_input,
|
15 |
outputs="text",
|
16 |
-
title="
|
17 |
description="Upload an audio here and get a bullet-point summary of its content.",
|
18 |
theme="Monochrome",
|
19 |
live=True,
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import BartTokenizer, BartForConditionalGeneration
|
3 |
import whisper
|
4 |
|
5 |
+
# Initialize the BART model and tokenizer
|
6 |
+
MODEL_NAME = "facebook/bart-large-cnn"
|
7 |
+
model = BartForConditionalGeneration.from_pretrained(MODEL_NAME)
|
8 |
+
tokenizer = BartTokenizer.from_pretrained(MODEL_NAME)
|
9 |
|
10 |
+
def convert_and_summarize(audio_path: str) -> str:
|
11 |
+
# Convert audio to text
|
12 |
+
whisper_model = whisper.load_model("base")
|
13 |
+
result = whisper_model.transcribe(audio_path)
|
14 |
+
transcribed_text = result["text"]
|
15 |
+
|
16 |
+
# Summarize the transcribed text
|
17 |
+
inputs = tokenizer([transcribed_text], max_length=1024, truncation=True, return_tensors='pt')
|
18 |
+
summary_ids = model.generate(inputs['input_ids'])
|
19 |
+
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
20 |
+
|
21 |
+
return summary
|
22 |
+
|
23 |
+
audio_input = gr.inputs.Audio(type="filepath")
|
24 |
|
25 |
# Interface for Gradio
|
26 |
iface = gr.Interface(
|
27 |
+
fn=convert_and_summarize,
|
28 |
inputs=audio_input,
|
29 |
outputs="text",
|
30 |
+
title="Audio-to-Summarized-Text",
|
31 |
description="Upload an audio here and get a bullet-point summary of its content.",
|
32 |
theme="Monochrome",
|
33 |
live=True,
|