DEV Community

Cover image for How to Record Audio in Python: Automatically Detect Speech and Silence
Abhinav Anand
Abhinav Anand

Posted on

18

How to Record Audio in Python: Automatically Detect Speech and Silence

Recording audio only when someone is speaking is a powerful feature that can be used in various applications, from voice-activated assistants to saving storage space by eliminating silent periods. In this tutorial, you'll learn how to write Python code that starts recording when it detects speech and stops when silence is detected.

Prerequisites

Before diving in, ensure you have the following:

  • Python 3.x installed on your system.
  • Basic knowledge of Python.
  • Familiarity with Python libraries like pyaudio, numpy, and webrtcvad.

Step 1: Install Required Libraries 📦

We’ll be using the following libraries:

  • pyaudio: For capturing audio from your microphone.
  • webrtcvad: For voice activity detection.
  • numpy: For handling audio data.

You can install them using pip:

pip install pyaudio webrtcvad numpy
Enter fullscreen mode Exit fullscreen mode

Step 2: Setting Up Audio Stream 🎧

First, let’s set up the audio stream to capture audio input from your microphone.

import pyaudio

# Audio configuration
FORMAT = pyaudio.paInt16
CHANNELS = 1
RATE = 16000
CHUNK = 1024

# Initialize PyAudio
audio = pyaudio.PyAudio()

# Open stream
stream = audio.open(format=FORMAT,
                    channels=CHANNELS,
                    rate=RATE,
                    input=True,
                    frames_per_buffer=CHUNK)
Enter fullscreen mode Exit fullscreen mode

Step 3: Implementing Voice Activity Detection (VAD) 🎤

We’ll use the webrtcvad library to detect when someone is speaking. The library can classify audio frames as speech or non-speech.

import webrtcvad

# Initialize VAD
vad = webrtcvad.Vad()
vad.set_mode(1)  # 0: Aggressive filtering, 3: Less aggressive

def is_speech(frame, sample_rate):
    return vad.is_speech(frame, sample_rate)
Enter fullscreen mode Exit fullscreen mode

Step 4: Capturing and Processing Audio Frames 🎼

Now, let's continuously capture audio frames and check if they contain speech.

def record_audio():
    frames = []
    recording = False

    print("Listening for speech...")

    while True:
        frame = stream.read(CHUNK)

        if is_speech(frame, RATE):
            if not recording:
                print("Recording started.")
                recording = True
            frames.append(frame)
        else:
            if recording:
                print("Silence detected, stopping recording.")
                break

    # Stop and close the stream
    stream.stop_stream()
    stream.close()
    audio.terminate()

    return frames
Enter fullscreen mode Exit fullscreen mode

Step 5: Saving the Recorded Audio 💾

Finally, let’s save the recorded audio to a .wav file.

import wave

def save_audio(frames, filename="output.wav"):
    wf = wave.open(filename, 'wb')
    wf.setnchannels(CHANNELS)
    wf.setsampwidth(audio.get_sample_size(FORMAT))
    wf.setframerate(RATE)
    wf.writeframes(b''.join(frames))
    wf.close()

# Example usage
frames = record_audio()
save_audio(frames)
print("Audio saved as output.wav")
Enter fullscreen mode Exit fullscreen mode

Conclusion 🎉

With just a few lines of code, you’ve implemented a Python program that detects speech and records only the speaking portions, ignoring silence. This technique is especially useful for creating efficient voice-activated systems.

Feel free to experiment with the VAD aggressiveness and audio settings to suit your specific needs. Happy coding! 👩‍💻👨‍💻


Image of Datadog

The Future of AI, LLMs, and Observability on Google Cloud

Datadog sat down with Google’s Director of AI to discuss the current and future states of AI, ML, and LLMs on Google Cloud. Discover 7 key insights for technical leaders, covering everything from upskilling teams to observability best practices

Learn More

Top comments (2)

Collapse
 
pablo_ff9cf126a8f4166f37a profile image
Pablo

hey, I got:


Error Traceback (most recent call last)
Cell In[10], line 72
69 wf.close()
71 # Example usage
---> 72 frames = record_audio()
73 save_audio(frames)
74 print("Audio saved as output.wav")

Cell In[10], line 43
40 while True:
41 frame = stream.read(CHUNK, exception_on_overflow=False)
---> 43 if is_speech(frame, RATE):
44 if not recording:
45 print("Recording started.")

Cell In[10], line 30
28 pcm_data = struct.unpack(f'{len(frame) // 2}h', frame) # Convert bytes to 16-bit samples
29 packed_pcm_data = struct.pack(f'{len(pcm_data)}h', *pcm_data) # Repack to ensure correct format
---> 30 return vad.is_speech(packed_pcm_data, sample_rate)

File c:\Users\pablo\Desktop\P - Proyectos en Curso\ai-curated-articles\ai_curator\Lib\site-packages\webrtcvad.py:37, in Vad.is_speech(self, buf, sample_rate, length)
33 if length * 2 > len(buf):
34 raise IndexError(
35 'buffer has %s frames, but length argument was %s' % (
36 int(len(buf) / 2.0), length))
---> 37 return _webrtcvad.process(self._vad, sample_rate, buf, length)

Error: Error while processing frame

Collapse
 
abhinowww profile image
Abhinav Anand

hey can you share the source code

Image of Docusign

🛠️ Bring your solution into Docusign. Reach over 1.6M customers.

Docusign is now extensible. Overcome challenges with disconnected products and inaccessible data by bringing your solutions into Docusign and publishing to 1.6M customers in the App Center.

Learn more