Join us and get access to thousands of tutorials and a community of expert Pythonistas.

Unlock This Lesson

This lesson is for members only. Join us and get access to thousands of tutorials and a community of expert Pythonistas.

Unlock This Lesson

python-sounddevice (Part 2)

00:00 python-sounddevice is going to record audio from your microphone and store it as a NumPy array. If you’d like to convert a NumPy array to a WAV file, you’ll want to use a module from SciPy to do so. Let’s go ahead and install SciPy to get started.

00:19 Going to the text editor, you’re going to import sounddevice as sd, and then from, and you’ll just want to import write. So before, when we were playing audio, python-sounddevice obtained a lot of information from the WAV file itself. Since you’re going to be creating a WAV file, you’ll need to set a sample rate, so using fs set the sample rate to 44100, and then a duration, so seconds, and let’s go 3. Now it’s time to create the recording object, so just go ahead and say something like myrecording and equal this to sd.rec() (record), and pass in an integer of seconds times your sample rate.

01:13 This will give you how many samples are going to be captured.

01:19 And then for samplerate, just pass in that value again, and channels=2. All right, let me hide this EXPLORER so we can see a little bit better.

01:31 You’ll want to wait until the recording is complete before you write anything, so throw in sd.wait(), and then go ahead and write(), and we’ll just call this 'output.wav'. And you’ll want to pass it in your sample rate, and then that actual recording object.

01:49 All right, let’s test this out. So down in the terminal, let’s run python_sounds. Okay. On my other screen, I got a notice to get access to the microphone, so I’m just going to say OK.

02:02 And we’ll see if this records! Okay, it seemed like it was about three seconds, so let’s take a look at the EXPLORER again. And there’s output.wav.

02:13 Let’s use one of the libraries from before and see if this worked. I’m just going to comment out everything there, and from playsound import playsound,

02:27 and then just grab playsound('output.wav').

02:34 Let’s run that again and see if it worked. “And we’ll see if this records.” Awesome! So that went to the microphone on my laptop, but you can see that we were able to record something using python-sounddevice.

02:47 Pretty cool! So, python-sounddevice is a good high-level recording option to use. In the next video, we’re going to take a look at pyaudio, which you may remember was a little bit more complex, but a very powerful library for dealing with audio in Python.

Pygator on March 1, 2020

channels=2 means what exactly?

Joe Tatusko RP Team on March 2, 2020

Stereo! Two channels will get you one for a left and right speaker, while one channel produces a mono output.

Hi, I tried this code on my mac,

fs = 44100
seconds = 3

my_recording = sd.rec(int(seconds * fs), samplerate=fs, channels=2)
write('output.wav', fs, my_recording)


but it gives me

raise PortAudioError(errormsg, err)
sounddevice.PortAudioError: Error opening InputStream: Invalid number of channels [PaErrorCode -9998]

Then I tried to change channels to 1, it works, but it didn’t record my sound. You know how to fix this issue? Thanks.

Joe Tatusko RP Team on March 26, 2020

Hi TJ,

Have you been able to record sound with any of the other scripts in this course yet? MacOS has become a lot more strict with microphone permissions recently and it may be playing a role. Try running it with two channels but from Terminal and see if that helps you out (I have to do with with some programs to prompt the “Allow ______ to use the microphone?” request)

Hi, Joe

I tried to run the sounddevice in my regular terminal, and it still didn’t work. I also tried to install pyaudio by pipenv, but it gives me the new error message.

Installing dependencies from Pipfile.lock (611ef3)
An error occurred while installing pyaudio==0.2.11 --hash=sha256:0d92f6a294565260a282f7c9a0b0d309fc8cc988b5ee5b50645634ab9e2da7f7 --hash=sha256:259bb9c1363be895b4f9a97e320a6017dd06bc540728c1a04eb4a7b6fe75035b --hash=sha256:2a19bdb8ec1445b4f3e4b7b109e0e4cec1fd1f1ce588592aeb6db0b58d4fb3b0 --hash=sha256:51b558d1b28c68437b53218279110db44f69f3f5dd3d81859f569a4a96962bdc --hash=sha256:589bfad2c615dd4b5d3757e763019c42ab82f06fba5cae64ec02fd7f5ae407ed --hash=sha256:8f89075b4844ea94dde0c951c2937581c989fabd4df09bfd3f075035f50955df --hash=sha256:93bfde30e0b64e63a46f2fd77e85c41fd51182a4a3413d9edfaf9ffaa26efb74 --hash=sha256:cf1543ba50bd44ac0d0ab5c035bb9c3127eb76047ff12235149d9adf86f532b6 --hash=sha256:f78d543a98b730e64621ebf7f3e2868a79ade0a373882ef51c0293455ffa8e6e! Will try again.

achk on Aug. 19, 2020

myrecording = sd.rec(int(seconds * fs), samplerate=fs, channels=2) AttributeError: ‘module’ object has no attribute ‘rec’

achk on Aug. 19, 2020

sounddevice.PortAudioError: Error opening InputStream: Invalid number of channels [PaErrorCode -9998]

Become a Member to join the conversation.