Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import speech_recognition as sr
|
3 |
+
from pocketsphinx import pocketsphinx, Jsgf, FsgModel
|
4 |
+
import requests
|
5 |
+
import os
|
6 |
+
|
7 |
+
|
8 |
+
st.title("Speech to text recognition")
|
9 |
+
|
10 |
+
# st.markdown("## Here we use pocketsphinx model for automatic speech recognition")
|
11 |
+
|
12 |
+
audio = st.file_uploader(label = "Upload your audio file here in .wav format")
|
13 |
+
|
14 |
+
|
15 |
+
# audio_file = '/Users/kapilgupta/Downloads/audio/videoplayback.wav'
|
16 |
+
text_filename = "./subfolder/text_file"
|
17 |
+
|
18 |
+
|
19 |
+
language_model = './language-model.lm.bin'
|
20 |
+
acoustic_model = './acoustic-model'
|
21 |
+
pronunciation_dict = './pronounciation-dictionary.dict'
|
22 |
+
|
23 |
+
@st.cache
|
24 |
+
def model(audio, text_filename):
|
25 |
+
framerate = 100
|
26 |
+
config = pocketsphinx.Config()
|
27 |
+
config.set_string('-hmm', acoustic_model)
|
28 |
+
config.set_string('-lm', language_model)
|
29 |
+
config.set_string('-dict', pronunciation_dict)
|
30 |
+
decoder = pocketsphinx.Decoder(config)
|
31 |
+
|
32 |
+
def recognize_sphinx(audio, show_all=True):
|
33 |
+
decoder.start_utt()
|
34 |
+
decoder.process_raw(audio.get_raw_data(), False, True)
|
35 |
+
decoder.end_utt()
|
36 |
+
hypothesis = decoder.hyp()
|
37 |
+
return decoder, hypothesis.hypstr
|
38 |
+
|
39 |
+
# Create a Recognizer instance
|
40 |
+
r = sr.Recognizer()
|
41 |
+
|
42 |
+
# Set the recognize_sphinx() function as the speech recognition method
|
43 |
+
r.recognize_sphinx = recognize_sphinx
|
44 |
+
|
45 |
+
with sr.AudioFile(audio) as source:
|
46 |
+
audio = r.record(source)
|
47 |
+
sample_rate = audio.sample_rate
|
48 |
+
decoder, recognized_text = r.recognize_sphinx(audio, show_all=True)
|
49 |
+
|
50 |
+
|
51 |
+
with open(text_filename, 'w') as text_file:
|
52 |
+
for seg in decoder.seg():
|
53 |
+
segment_info = (seg.word, seg.start_frame/sample_rate, seg.end_frame/sample_rate)
|
54 |
+
text_file.write(str(segment_info) + "\n")
|
55 |
+
return recognized_text
|
56 |
+
|
57 |
+
|
58 |
+
if audio is not None:
|
59 |
+
with st.spinner("code is at Working! "):
|
60 |
+
segment_info = model(audio, text_filename)
|
61 |
+
st.write(segment_info)
|
62 |
+
st.balloons()
|
63 |
+
else:
|
64 |
+
st.write("Upload an audio")
|