Play and record sound using pyaudio simultaneously

Question

I'm trying to create a program to talk back at once. I can't seem to get it to work. Some websites say use numpy arrays but I don't know how.

import pyaudio
import wave
import time
import multiprocessing as mp
import pyaudio
import numpy as np
import sounddevice as sd

fs = 44100
FORMAT = pyaudio.paInt16
CHANNELS = 2
RATE = 44100
CHUNK = 1024
audio = pyaudio.PyAudio()
RECORD_SECONDS = 5
stream = audio.open(format=FORMAT, channels=CHANNELS,
                rate=RATE, input=True,
                frames_per_buffer=CHUNK)
myarray = []
for i in range(0, int(RATE / CHUNK * RECORD_SECONDS)):
    data = stream.read(CHUNK)
    myarray.append(data)

    myrecording = sd.play(myarray, fs, channels=2)

Traceback (most recent call last): File "SoundTest.py", line 24, in myrecording = sd.play(myarray, fs, channels=2) line 2170, in check_data dtype = _check_dtype(data.dtype) File "/home/lordvile/.local/lib/python2.7/site-packages/sounddevice.py", line 2316, in _check_dtype raise TypeError('Unsupported data type: ' + repr(dtype)) TypeError: Unsupported data type: 'string32768'

@Michael I really have no idea. This is the error. I removed a few lines since it wont fit...Traceback (most recent call last): File "SoundTest.py", line 24, in myrecording = sd.play(myarray, fs, channels=2) line 2170, in check_data dtype = _check_dtype(data.dtype) File "/home/lordvile/.local/lib/python2.7/site-packages/sounddevice.py", line 2316, in _check_dtype raise TypeError('Unsupported data type: ' + repr(dtype)) TypeError: Unsupported data type: 'string32768' — Subhaditya Mukherjee, Aug 10 '16 at 16:33
@SubhadityaMukherjee Update your question with your Tracebak please. — julian salas, Aug 10 '16 at 16:38
Looks like `sounddevice` does not like `pyaudio`s data. Have you checked that the format is compatible? — handle, Aug 10 '16 at 17:25
@handle It appears so.. No I havent checked actually. To be honest im still in school and I dont think we will be doing numpy at all.. I dont understand how do I create the list with the audio data? It doesnt seem to be working. And thanks for helping.. — Subhaditya Mukherjee, Aug 11 '16 at 06:04
You shouldn't mix `PyAudio` and the `sounddevice` module. Use either one or the other. — Matthias, Aug 25 '16 at 13:32

score 3 · Answer 1 · answered May 20 '20 at 21:26

Below are examples for both 'PyAudio' and 'Sounddevice' for simultaneous recording and playback. The most important thing for me was to explicitly set the input and output devices to a valid value. Hence, I included printing for sound devices for both examples.

Sounddevice:

Works like a charm compared to PyAudio. At least for me, sounddevice offers a stable interface for simultaneous audio recording and playback. Though there are a few console outputs complaining about a input/output over-/underflow, the record/playback regenerates and eventually I get a fluently working 'echo'.

import sounddevice as sd

fs = 44100

def callback(indata, outdata, frames, time, status):
    if status:
        print(status)
    outdata[:] = indata

print(sd.query_devices())

try:
    with sd.Stream(device=(0,1), samplerate=fs, dtype='float32', latency=None, channels=2, callback=callback):
        input()
except KeyboardInterrupt:
    pass

PyAudio:

This code should(?) work. For me it does not work consistently. It already worked for about a minute once but usually I get the Error: "OSError: [Errno -9981] Input overflowed". I don't know why it always throws this error, but feel free to try it out yourself.

import pyaudio
import json

FORMAT = pyaudio.paInt16
CHANNELS = 1
CHUNK = 1024
RATE = 44100

audio = pyaudio.PyAudio()

for i in range(audio.get_device_count()):
    print(json.dumps(audio.get_device_info_by_index(i), indent=2))

stream = audio.open(format              = FORMAT,
                    channels            = CHANNELS,
                    rate                = RATE,
                    input               = True,
                    output              = True,
                    input_device_index  = 0,
                    output_device_index = 1,
                    frames_per_buffer   = CHUNK)

try:
    frames = []
    print("* echoing")
    print("Press CTRL+C to stop")
    while True:
        data = stream.read(CHUNK)
        frames.append(data)
        if len(frames)>0:
            stream.write(frames.pop(0),CHUNK)
    print("* done echoing")
except KeyboardInterrupt:
    stream.stop_stream()
    stream.close()
    audio.terminate()

For me who uses PyAudio, explicitly specifying the input device index and output device index solves the problem — goodahn, Jul 07 '20 at 15:52
The two examples compare `sounddevice` using "callback" mode with `pyaudio` using "blocking" mode. Both modules can use both modes, so it would be more meaningful to compare them using the same mode. — Matthias, Apr 16 '21 at 08:19
Note that `pyaudio` has the `exception_on_overflow` parameter for `read()` (see https://people.csail.mit.edu/hubert/pyaudio/docs/#pyaudio.Stream.read), which is `True` by default. If you set this to `False`, you can avoid raising the mentioned exception (but I don't recommend ignoring those over- and underflows!). — Matthias, Apr 16 '21 at 08:22

score 2 · Answer 2 · answered Aug 30 '16 at 15:21

Not sure if you have requirement to use pyaudio, but here is an example using sounddevice.

Below example records audio from microphone for #seconds as per variable duration which you can modify as per your requirements.

Same contents are played back using standard audio output (speakers).

Play and record sound using pyaudio simultaneously

2 Answers2

Linked