Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,96 +1,33 @@
|
|
1 |
-
# import torchaudio
|
2 |
-
# import gradio as gr
|
3 |
-
# from transformers import pipeline
|
4 |
-
# from gtts import gTTS
|
5 |
-
# import tempfile
|
6 |
-
# import pygame
|
7 |
-
# import time
|
8 |
-
|
9 |
-
# # Initialize the speech-to-text transcriber
|
10 |
-
# transcriber = pipeline("automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-english")
|
11 |
-
|
12 |
-
# # Load the pre-trained question answering model
|
13 |
-
# model_name = "AVISHKAARAM/avishkaarak-ekta-hindi"
|
14 |
-
# qa_model = pipeline("question-answering", model=model_name)
|
15 |
-
|
16 |
-
# def answer_question(context, question=None, audio=None):
|
17 |
-
# if audio is not None:
|
18 |
-
# text = transcriber(audio)
|
19 |
-
# question_text = text['text']
|
20 |
-
# else:
|
21 |
-
# question_text = question
|
22 |
-
|
23 |
-
# qa_result = qa_model(question=question_text, context=context)
|
24 |
-
# answer = qa_result["answer"]
|
25 |
-
|
26 |
-
# tts = gTTS(text=answer, lang='en')
|
27 |
-
# audio_path = tempfile.NamedTemporaryFile(suffix=".mp3").name
|
28 |
-
# tts.save(audio_path)
|
29 |
-
|
30 |
-
# return answer, audio_path
|
31 |
-
|
32 |
-
# def play_audio(audio_path):
|
33 |
-
# pygame.mixer.init()
|
34 |
-
# pygame.mixer.music.load(audio_path)
|
35 |
-
# pygame.mixer.music.play()
|
36 |
-
# while pygame.mixer.music.get_busy():
|
37 |
-
# time.sleep(0.1)
|
38 |
-
|
39 |
-
# # Define the Gradio interface
|
40 |
-
# context_input = gr.components.Textbox(label="Context")
|
41 |
-
# question_input = gr.components.Textbox(label="Question")
|
42 |
-
# audio_input = gr.components.Audio(source="microphone", type="filepath")
|
43 |
-
|
44 |
-
# output_text = gr.components.Textbox(label="Answer")
|
45 |
-
# output_audio = gr.components.Audio(label="Answer Audio", type="numpy")
|
46 |
-
|
47 |
-
# inter = gr.Interface(
|
48 |
-
# fn=answer_question,
|
49 |
-
# inputs=[context_input, question_input, audio_input],
|
50 |
-
# outputs=[output_text, output_audio],
|
51 |
-
# title="Question Answering",
|
52 |
-
# description="Enter a context and a question to get an answer. You can also upload an audio file with the question.",
|
53 |
-
# examples=[
|
54 |
-
# ["The capital of France is Paris.", "What is the capital of France?"],
|
55 |
-
# ["OpenAI is famous for developing GPT-3.", "What is OpenAI known for?"],
|
56 |
-
# ]
|
57 |
-
# )
|
58 |
-
# inter.launch()
|
59 |
-
|
60 |
-
|
61 |
-
import torchaudio
|
62 |
import gradio as gr
|
63 |
from transformers import pipeline
|
64 |
-
import
|
65 |
import tempfile
|
66 |
-
import time
|
67 |
|
68 |
# Initialize the speech-to-text transcriber
|
|
|
69 |
transcriber = pipeline("automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-english")
|
70 |
|
71 |
-
#
|
72 |
model_name = "AVISHKAARAM/avishkaarak-ekta-hindi"
|
73 |
qa_model = pipeline("question-answering", model=model_name)
|
74 |
|
75 |
-
# Initialize pyttsx3 TTS
|
76 |
-
engine = pyttsx3.init()
|
77 |
-
|
78 |
def answer_question(context, question=None, audio=None):
|
79 |
# Handle audio input
|
80 |
if audio is not None:
|
81 |
-
text
|
82 |
-
|
|
|
83 |
else:
|
84 |
question_text = question
|
85 |
|
86 |
-
# Generate the answer
|
87 |
qa_result = qa_model(question=question_text, context=context)
|
88 |
answer = qa_result["answer"]
|
89 |
|
90 |
-
# Convert answer to speech
|
|
|
91 |
audio_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
|
92 |
-
|
93 |
-
engine.runAndWait()
|
94 |
|
95 |
return answer, audio_path
|
96 |
|
@@ -116,4 +53,3 @@ inter = gr.Interface(
|
|
116 |
|
117 |
# Launch the Gradio interface
|
118 |
inter.launch()
|
119 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
from transformers import pipeline
|
3 |
+
from gtts import gTTS
|
4 |
import tempfile
|
|
|
5 |
|
6 |
# Initialize the speech-to-text transcriber
|
7 |
+
from transformers import pipeline
|
8 |
transcriber = pipeline("automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-english")
|
9 |
|
10 |
+
# Initialize the pre-trained question-answering model
|
11 |
model_name = "AVISHKAARAM/avishkaarak-ekta-hindi"
|
12 |
qa_model = pipeline("question-answering", model=model_name)
|
13 |
|
|
|
|
|
|
|
14 |
def answer_question(context, question=None, audio=None):
|
15 |
# Handle audio input
|
16 |
if audio is not None:
|
17 |
+
# Convert audio to text using transcriber
|
18 |
+
transcription_result = transcriber(audio)["text"]
|
19 |
+
question_text = transcription_result
|
20 |
else:
|
21 |
question_text = question
|
22 |
|
23 |
+
# Generate the answer using the QA model
|
24 |
qa_result = qa_model(question=question_text, context=context)
|
25 |
answer = qa_result["answer"]
|
26 |
|
27 |
+
# Convert the answer to speech using gTTS
|
28 |
+
tts = gTTS(text=answer, lang='en')
|
29 |
audio_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
|
30 |
+
tts.save(audio_path)
|
|
|
31 |
|
32 |
return answer, audio_path
|
33 |
|
|
|
53 |
|
54 |
# Launch the Gradio interface
|
55 |
inter.launch()
|
|