Update app.py
Browse files
app.py
CHANGED
|
@@ -1,8 +1,6 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import torch
|
| 3 |
from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification
|
| 4 |
-
import tempfile
|
| 5 |
-
import torchaudio
|
| 6 |
|
| 7 |
# Load Whisper for transcription
|
| 8 |
asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3")
|
|
@@ -15,21 +13,16 @@ grammar_pipeline = pipeline("text-classification", model=cola_model, tokenizer=c
|
|
| 15 |
# Load grammar correction model
|
| 16 |
correction_pipeline = pipeline("text2text-generation", model="vennify/t5-base-grammar-correction")
|
| 17 |
|
| 18 |
-
def process_audio(
|
| 19 |
-
#
|
| 20 |
-
|
| 21 |
-
tmp.write(audio_file.read())
|
| 22 |
-
tmp_path = tmp.name
|
| 23 |
|
| 24 |
-
#
|
| 25 |
-
transcription = asr_pipeline(tmp_path)["text"]
|
| 26 |
-
|
| 27 |
-
# Grammar Scoring
|
| 28 |
grammar_result = grammar_pipeline(transcription)[0]
|
| 29 |
score_label = grammar_result["label"]
|
| 30 |
score_confidence = grammar_result["score"]
|
| 31 |
|
| 32 |
-
#
|
| 33 |
corrected_text = correction_pipeline(transcription, max_length=128)[0]["generated_text"]
|
| 34 |
|
| 35 |
return transcription, f"{score_label} ({score_confidence:.2f})", corrected_text
|
|
@@ -37,14 +30,18 @@ def process_audio(audio_file):
|
|
| 37 |
# Gradio Interface
|
| 38 |
interface = gr.Interface(
|
| 39 |
fn=process_audio,
|
| 40 |
-
inputs=gr.Audio(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
outputs=[
|
| 42 |
-
gr.Textbox(label="Transcription"),
|
| 43 |
-
gr.Textbox(label="Grammar Score"),
|
| 44 |
-
gr.Textbox(label="
|
| 45 |
],
|
| 46 |
title="ποΈ Voice Grammar Scorer",
|
| 47 |
-
description="
|
| 48 |
)
|
| 49 |
|
| 50 |
if __name__ == "__main__":
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import torch
|
| 3 |
from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification
|
|
|
|
|
|
|
| 4 |
|
| 5 |
# Load Whisper for transcription
|
| 6 |
asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3")
|
|
|
|
| 13 |
# Load grammar correction model
|
| 14 |
correction_pipeline = pipeline("text2text-generation", model="vennify/t5-base-grammar-correction")
|
| 15 |
|
| 16 |
+
def process_audio(audio_path):
|
| 17 |
+
# Transcribe
|
| 18 |
+
transcription = asr_pipeline(audio_path)["text"]
|
|
|
|
|
|
|
| 19 |
|
| 20 |
+
# Score grammar
|
|
|
|
|
|
|
|
|
|
| 21 |
grammar_result = grammar_pipeline(transcription)[0]
|
| 22 |
score_label = grammar_result["label"]
|
| 23 |
score_confidence = grammar_result["score"]
|
| 24 |
|
| 25 |
+
# Suggest correction
|
| 26 |
corrected_text = correction_pipeline(transcription, max_length=128)[0]["generated_text"]
|
| 27 |
|
| 28 |
return transcription, f"{score_label} ({score_confidence:.2f})", corrected_text
|
|
|
|
| 30 |
# Gradio Interface
|
| 31 |
interface = gr.Interface(
|
| 32 |
fn=process_audio,
|
| 33 |
+
inputs=gr.Audio(
|
| 34 |
+
source="microphone", # enables both mic recording and upload
|
| 35 |
+
type="filepath",
|
| 36 |
+
label="π€ Record or Upload Audio (.wav)"
|
| 37 |
+
),
|
| 38 |
outputs=[
|
| 39 |
+
gr.Textbox(label="π Transcription"),
|
| 40 |
+
gr.Textbox(label="β
Grammar Score"),
|
| 41 |
+
gr.Textbox(label="βοΈ Suggested Correction")
|
| 42 |
],
|
| 43 |
title="ποΈ Voice Grammar Scorer",
|
| 44 |
+
description="Record or upload your voice (.wav). This app transcribes it, scores grammar, and suggests corrections."
|
| 45 |
)
|
| 46 |
|
| 47 |
if __name__ == "__main__":
|