Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Sign up
Appearance settings

Commit 04f192c

Browse files
Merge pull request avinashkranjan#1927 from srujana-16/Audio-Splitting
 Audio splitting script added - GSSOC'23
2 parents 7c96bba + 85b5868 commit 04f192c

File tree

3 files changed

+183
-0
lines changed

3 files changed

+183
-0
lines changed

‎Audio Splitting/README.md

Lines changed: 22 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,22 @@
1+
# Audio Splitting
2+
3+
This Python script allows you to split large audio files into smaller segments based on silence, making them suitable for transcription or further analysis.
4+
5+
## Setup instructions
6+
7+
- Download the repository and `cd` to the the folder by running the caommand `cd Audio\ splitter`
8+
- Run the script with the following command `python audio_split.py input_audio_file`
9+
- `input_audio_file` - Provide the path to the audio file you want to split.
10+
11+
## Output
12+
13+
The audio chunks will be saved in a new folder on your local system. Each audio chunk is approximately 10-15 seconds long which is best suitable for transcription.
14+
15+
## Author(s)
16+
17+
Srujana
18+
19+
## Disclaimers, if any
20+
21+
Use this section to mention if any particular disclaimer is required
22+

‎Audio Splitting/audio-split.py

Lines changed: 159 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,159 @@
1+
from pydub import AudioSegment
2+
from pydub.silence import split_on_silence
3+
import os
4+
import collections
5+
import contextlib
6+
import sys
7+
import wave
8+
import os
9+
import webrtcvad
10+
11+
def read_wave(path):
12+
13+
with contextlib.closing(wave.open(path, 'rb')) as wf:
14+
num_channels = wf.getnchannels()
15+
assert num_channels == 1
16+
sample_width = wf.getsampwidth()
17+
assert sample_width == 2
18+
sample_rate = wf.getframerate()
19+
assert sample_rate in (8000, 16000, 32000, 48000)
20+
pcm_data = wf.readframes(wf.getnframes())
21+
return pcm_data, sample_rate
22+
23+
24+
def write_wave(path, audio, sample_rate):
25+
26+
with contextlib.closing(wave.open(path, 'wb')) as wf:
27+
wf.setnchannels(1)
28+
wf.setsampwidth(2)
29+
wf.setframerate(sample_rate)
30+
wf.writeframes(audio)
31+
frames = wf.getnframes()
32+
return frames / float(sample_rate)
33+
34+
35+
class Frame(object):
36+
37+
def __init__(self, bytes, timestamp, duration):
38+
self.bytes = bytes
39+
self.timestamp = timestamp
40+
self.duration = duration
41+
42+
43+
def frame_generator(frame_duration_ms, audio, sample_rate):
44+
45+
n = int(sample_rate * (frame_duration_ms / 1000.0) * 2)
46+
offset = 0
47+
timestamp = 0.0
48+
duration = (float(n) / sample_rate) / 2.0
49+
while offset + n < len(audio):
50+
yield Frame(audio[offset:offset + n], timestamp, duration)
51+
timestamp += duration
52+
offset += n
53+
54+
55+
def vad_collector(sample_rate, frame_duration_ms,
56+
padding_duration_ms, vad, frames):
57+
58+
num_padding_frames = int(padding_duration_ms / frame_duration_ms)
59+
ring_buffer = collections.deque(maxlen=num_padding_frames)
60+
triggered = False
61+
62+
voiced_frames = []
63+
for frame in frames:
64+
is_speech = vad.is_speech(frame.bytes, sample_rate)
65+
66+
67+
if not triggered:
68+
ring_buffer.append((frame, is_speech))
69+
num_voiced = len([f for f, speech in ring_buffer if speech])
70+
71+
if num_voiced > 0.9 * ring_buffer.maxlen:
72+
triggered = True
73+
74+
75+
for f, s in ring_buffer:
76+
voiced_frames.append(f)
77+
ring_buffer.clear()
78+
else:
79+
80+
voiced_frames.append(frame)
81+
ring_buffer.append((frame, is_speech))
82+
num_unvoiced = len([f for f, speech in ring_buffer if not speech])
83+
84+
if num_unvoiced > 0.9 * ring_buffer.maxlen:
85+
86+
triggered = False
87+
yield b''.join([f.bytes for f in voiced_frames])
88+
ring_buffer.clear()
89+
voiced_frames = []
90+
if triggered:
91+
pass
92+
93+
if voiced_frames:
94+
yield b''.join([f.bytes for f in voiced_frames])
95+
96+
path = "./frontend/speech-transcription-app/public/Original data"
97+
if not os.path.exists(path):
98+
os.makedirs(path)
99+
print("Output folder created")
100+
else:
101+
print("Output folder already present")
102+
sys.exit()
103+
104+
def folder(path):
105+
if not os.path.exists(path):
106+
os.makedirs(path)
107+
print("Output folder created")
108+
else:
109+
print("Output folder already present")
110+
111+
path = "./frontend/speech-transcription-app/public/Original data"
112+
folder(path)
113+
path = "./main/save"
114+
folder(path)
115+
path = "./main/discard"
116+
folder(path)
117+
118+
file_name= "./main/mod_1.wav"
119+
op_path= "./frontend/speech-transcription-app/public/Original data/audio_chunks"
120+
121+
122+
123+
def main(file_name,op_path):
124+
125+
if os.path.isdir(op_path):
126+
print("Output folder already present")
127+
else:
128+
os.mkdir(op_path)
129+
print("Output folder created")
130+
131+
audio, sample_rate = read_wave(file_name)
132+
vad = webrtcvad.Vad(2)
133+
frames = frame_generator(30, audio, sample_rate)
134+
segments = vad_collector(sample_rate, 30, 300, vad, frames)
135+
136+
for i, segment in enumerate(segments):
137+
path = op_path+'/'+'chunk%004d.wav' % (i+1,)
138+
print(' Writing %s' %(path,))
139+
write_wave(path, segment, sample_rate )
140+
141+
142+
143+
# sys.argv[1]
144+
145+
# sys.argv[2]
146+
file_name= "./main/mod_1.wav"
147+
op_path= "./frontend/speech-transcription-app/public/Original data/audio_chunks"
148+
main(file_name,op_path)
149+
150+
print("Audio Splitting Done")
151+
152+
153+
154+
155+
156+
157+
158+
159+

‎SCRIPTS.md

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -103,4 +103,6 @@
103103
| 101\. |Story Generator | This script randomly selects a character, setting, action, and conclusion from predefined lists and generates a story using these elements. | [Take Me](./Story_Generator/story_generator.py) | [Shraddha Singh](https://github.com/shraddha761)
104104
| 102\. | Scientific Calculator | This python program will help to solve the hardest problems of trigonometry, exponents functions, logarithmic functions, etc. in very easy way. This program has a best User interface of the scientific calculator and provide a best user experience | [Take Me](./Scientific_Calculator) | [Avdhesh Varshney](https://github.com/Avdhesh-Varshney)
105105
| 103\. | Chess Game | Chess Game is created using PYTHON script that provides user a interactive platform to play chess game. This program will create a interactive GUI of chess board for the user. Directly started in the terminal. | [Take Me](./Chess-Game/) | [Avdhesh Varshney](https:github.com/Avdhesh-Varshney)
106+
| 104\. | Audio splitting script | This Python script allows you to split large audio files into smaller segments based on silence, making them suitable for transcription or further analysis. | [Take Me](https://github.com/avinashkranjan/Amazing-Python-Scripts/tree/master/Audio\Splitting) | [Srujana Vanka](https://github.com/srujana-16) |
107+
106108

0 commit comments

Comments
(0)

AltStyle によって変換されたページ (->オリジナル) /