Is there a way using python to input a mp3/audio file to a microphone input?

Question

My main purpose is to change the pitch of the my voice and then input it to a voice room/voice call say like a zoom call or a hangouts meeting without playing it back to me.

I found 2 questions on a similar topic: 1. Playing mp3 file through microphone with python 2. How to play MP3 files into the microphone input jQuery

But these ones do not answer the question appropriately.

Something similar to this: https://github.com/jremmons/pyfakewebcam/blob/master/pyfakewebcam/pyfakewebcam.py

but for microphones?

Answer 1

You can use sounddevice, just tested it on Windows 10

you can create a stream and read data while as it records, naturally you will have a latency proportional to the size of the chunk you read.

This example records 6.4 seconds of audio at 16kHz. After finished you call the method stop. Each chunk in this example has 1024 samples corresponding to 64 milliseconds.

micStream = sd.InputStream(samplerate=16000, blocksize=1024, channels=1, dtype='float32')
micStream.start();
CHUNK_SIZE = 1024
NCHUNKS = 100
wavData = np.zeros(CHUNK_SIZE*NCHUNKS)
for i in range(0, NCHUNKS):
    chunkData, overflowed = micStream.read(CHUNK_SIZE);
    wavData[i*chunk:(i+1)*chunk] = np.mean(chunkData, axis=1) # ensure to be mono
micStream.stop()
plt.plot(wavData)

Answer 2

Required Module: pyaudio

install it through pip:

pip install pyaudio

NOTE: I've faced issues while trying to install pyaudio module in python 3.8 version but have read a lot of articles that with python 3.5 you won't face any issues.

If you want to save the audio recorded through microphone you can follow https://gist.github.com/mabdrabo/8678538 (to have .mp3 extension modify the variable WAVE_OUTPUT_FILENAME in the code to WAVE_OUTPUT_FILENAME = "file.mp3")

You can also use the SpeechRecognition module. follow this link https://www.youtube.com/watch?v=x8xjj6cR9Nc where he has created a python speech assistance (like google assistance) in a 30min video from scratch and github link to the code in video description.