Playing mp3 file through microphone with python

Question

Is there a way using python (and not any external software) to play a mp3 file like a microphone input?

For example, I have a mp3 file and with a python script it would play it through my mic so other in a voice room would hear it. As I say it is just an example.

Of course, I have done some research. I found out that I can use a software to create a virtual device and do few things to have the result. But my point is if it is possible without installing software but with some kind of python script?

Pleurisy

No no, I didn't mean speakers. My question is very similar to this one: #28918579 , the only difference is that I would like to have a solution in python. – Pleurisy

Then, try to re-word the Q somehow or add explanation, it's comfusing. You see, microphone and microphone input are two different things. I thought you wish to use microphone instead of a speaker, which can be done, but is inadvisable and will ruin the mic, and I don't see why would anyone do that. I'll answer your Q below, in an edit. – Diaphanous

Are you saying, you want to record a sound via one microphone and play on different speakers in a multiplayer game? Or is it you want to play prerecorded sound (mp3) on different speakers simultaneously? Your use of word microphone is throwing me off to understand question. – Synchrocyclotron

I don't think stereo mix or virtual cable is the right answer, though they could work. I think what you are trying to find is a way to mannually control the input of the mic. Am I right? – Finke

Answer 1

If you meant how to play MP3 using Python, well, this is a broad question.

Is it possible, without any dependencies, yes it is, but it is not worth it. Well, playing uncompressed audio is, but MP3, well, I'll explain below.

To play raw audio data from Python without installing pyaudio or pygame or similar, you first have to know the platform on which your script will be run.

Then implement a nice set of functions for choosing an audio device, setting up properties like sample rate, bit rate, mono/stereo..., feeding the stream to audio card and stopping the playback.

It is not hard, but to do it you have to use ctypes on Windows, PyObjC on Mac and Linux is special case as it supports many audio systems (probably use sockets to connect to PulseAudio or pipe to some process like aplay/paplay/mpeg123... or exploit gstreamer.).

But why go through all this just to avoid dependencies, when you have nice libraries out there with simple interfaces to access and use audio devices.

PyAudio is great one.

Well, that is your concern.

But, playing MP3 without external libraries, in real time, from pure Python, well, it's not exactly impossible, but it is very hard to achieve, and as far as I know nobody even tried doing it.

There is pure Python MP3 decoder implementation, but it is 10 times slower than necessary for real-time audio playback. It can be optimized for nearly full speed, but nobody is interested in doing so.

It has mostly educational value and it is used in cases where you do not need real-time speed.

This is what you should do:

Install pygame and use it to play MP3 directly

or:

Install PyAudio and some library that decodes Mp3, there are quite a few of them on pypi.python.org, and use it to decode the MP3 and feed the output to PyAudio.

There are some more possibilities, including pymedia, but I consider these the easiest solutions.

Okay, as we clarified what is really you need here is the answer.

I will leave first answer intact as you need that part too.

Now, you want to play audio to the recording stream, so that any application recording the audio input records the stuff that you are playing.

On Windows, this is called stereo mix and can be found in Volume Control, under audio input.

You choose stereo mix as your default input. Now, when you open an recording app which doesn't select itsown input channel, but uses the selected one (e.g. Skype) , it will record all coming out of your speakers and coming into your mic/line in.

I am not 100% sure whether this option will appear on all Windows or it is a feature of an audio card you have.

I am positive that Creative and Realtek audio cards supports it.

So, research this.

To select that option from Python, you have to connect to winmm.dll using ctypes and call the appropriate function. I do not know which one and with what arguments.

If this option is not present in volume control, there is nothing for it but to install a virtual audio card to do the loopback for you.

There might be such a software that comes packaged in as library so that you can use it from Python or whatever.

On Linux this should be easy using Pulseaudio. I do not know how, but I know that you can do it, redirect the streams etc. There is tutorial out there somewhere.

Then you can call that command from Python, to set to this and reset back to normal.

On Mac, well, I really have no idea, but it should be possible.

If you want your MP3 to be played only to the recording stream, and not on your speakers at all, well on Windows, you will not be able to do that without a loopback audio device.

On Linux, I am sure you will be able to do it, and on Mac it should be possible, but how is the Q.

I currently have no time to sniff around libraries etc. to provide you with some useful code, so you will have to do it yourself. But I hope my directions will help you.

Diaphanous

No I didn't mean playing mp3 file like that. I mean playing an mp3 file through microphone so for example in a multiplayer game I can play a music for others through my microphone. – Pleurisy

Answer 2

It is possible but it isn't 100% in python as it requires the installation of other software. (Also from what I know this specific answer only works on Windows, but it should be similar on Linux with PulseAudio instead of VB-Audio Cable, but I'm not a daily Linux user so I don't know.)

First download: https://www.vb-audio.com/Cable/, this will create a "Virtual Audio Cable" where programs can play music to the input device (What looks like a speaker) and it'll pipe it to the output device (What looks like a microphone).

Then run this command in cmd: pip install pygame==2.0.0.dev8 (or py -m pip install pygame==2.0.0.dev8, depending on your installation of python) [Also the reason it's the dev version is that it requires some functions only in sdl2, whereas the main branch uses sdl1)

Then:

>>> from pygame._sdl2 import get_num_audio_devices, get_audio_device_name #Get playback device names
>>> from pygame import mixer #Playing sound
>>> mixer.init() #Initialize the mixer, this will allow the next command to work
>>> [get_audio_device_name(x, 0).decode() for x in range(get_num_audio_devices(0))] #Returns playback devices
['Headphones (Oculus Virtual Audio Device)', 'MONITOR (2- NVIDIA High Definition Audio)', 'Speakers (High Definition Audio Device)', 'Speakers (NVIDIA RTX Voice)', 'CABLE Input (VB-Audio Virtual Cable)']
>>> mixer.quit() #Quit the mixer as it's initialized on your main playback device
>>> mixer.init(devicename='CABLE Input (VB-Audio Virtual Cable)') #Initialize it with the correct device
>>> mixer.music.load("Megalovania.mp3") #Load the mp3
>>> mixer.music.play() #Play it

To stop the music do: mixer.music.stop()

Also, the music doesn't play through your speakers, so you're going to have another python script or thread running that handles playing it through your speakers. (Also if you want it to play on a button press I recommend using the python library keyboard, the GitHub documentation is really good and you should be able to figure it out on your own.)

PS: This took me a while to figure out, your welcome.

PPS: I'm still trying to figure out a way to pipe your own mic through there as well since this method will obviously not pipe your real microphone in too, but looking into the source code of pygame is making my head hurt due to it all being written in C.

Hocus

Just use pyaudio instead of pygame and you can do wonders with streams and what goes where. If the virtual device does not allow multiple streams to be opened (one for recording from mic, and one for playing the MP3) from the same process then a drawback will be that the music and recording will have to be mixed together manually. But it shouldn't be so. Also MP3 decoding will have to be performed separately, but if need be, SDL and portaudio can work together just fine, although having whole pygame+pyaudio just for that makes no sense to me. – Diaphanous

Answer 3

Just an update on @PyPylia's answer for the benefit of anyone who struggled to implement this like I did.

Current Package Version: pygame 2.1.2 (SDL 2.0.18, Python 3.10.6)

Tested Systems: Windows 10 (21H2 - 19044.1288), (Should be the same process on Mac but this is untested as of now...)

First, you'll need to download the VB-Cable Virtual Mic Driver for your respective platform and install it. This provides us with a virtual mic that'll allow us to pass audio we play on our machine as a microphone input when using a video calling software (Google Meet, Microsoft Teams, Zoom). After that, it's all handled through the pygame module's audio package.

To get the audio device list:

from pygame import mixer, _sdl2 as devicer

mixer.init() # Initialize the mixer, this will allow the next command to work

# Returns playback devices, Boolean value determines whether they are Input or Output devices.
print("Inputs:", devicer.audio.get_audio_device_names(True))
print("Outputs:", devicer.audio.get_audio_device_names(False))

mixer.quit() # Quit the mixer as it's initialized on your main playback device

For example, My device returns:

Inputs: ['Microphone (High Definition Audio Device)', 'CABLE Output (VB-Audio Virtual Cable)']
Outputs: ['Speakers (High Definition Audio Device)', 'CABLE Input (VB-Audio Virtual Cable)']

Then, to playback the audio:

import time
from pygame import mixer

mixer.init(devicename = 'CABLE Input (VB-Audio Virtual Cable)') # Initialize it with the correct device
mixer.music.load("Toby Fox - Megalovania.mp3") # Load the mp3
mixer.music.play() # Play it

while mixer.music.get_busy():  # wait for music to finish playing
    time.sleep(1)

If you wish to play multiple tracks back to back, add the following code segments to the while loop above:

...

else:
    mixer.music.unload() # Unload the mp3 to free up system resources

mixer.music.load("Sleeping at Last - Saturn.wav") # Load the wav

...

Then, on the other end, inside the relevant software, just change the microphone input from the default to CABLE Output (VB-Audio Virtual Cable) to have those on the other end hear the audio from the source.

If you're using a newer version of the package and some of the listed methods don't seem to work because of an AttributeError: module 'pygame' has no attribute {method_name}, use pyup and search for the method in question, to see if there have been any changes to how the method is invoked. This was the main reason @PyPylia's code snippet no longer works unless you use an older version of pygame.

Aquatic

Thank you for updating my answer. I wrote this quite a while ago, and the libraries have updated now. Additionally, my code quality was worse back then, so thank you for writing this in such a clean and nice to read way. – Hocus

I tried following these steps, but I keep getting the following error: pygame.error: No such device. Edit: I am sure the string is correct because I get it from the list returned from get_audio_device_names. – Ausgleich

@Ausgleich Could you message me with your code snippet and the outputs that you get? – Aquatic

Answer 4

If you want to play an audio file in local directory, you may follow this flow.

#!/usr/bin/env python

import pyaudio
import socket
import sys
import os

CHUNK = 4096

output = os.path.join(BASE_DIR, "speech.wav") #WAV format Output file  name
wf = wave.open(output, 'rb')

p = pyaudio.PyAudio()

stream = p.open(format=p.get_format_from_width(wf.getsampwidth()), 
    channels=wf.getnchannels(), 
    rate=wf.getframerate(), 
    output=True)

try:
    while True:  
        data = wf.readframes(CHUNK)
        stream.write(data)  
        
except KeyboardInterrupt:
    pass

print('Shutting down')
s.close()
stream.close()
audio.terminate()

Trinary