PulseAudio: How to output the audio from PulseAudio to a wave file

I'm using PulseAudio to do some tests on our Call Center platform on Linux.

I'm using command python -m playwright codegen 'https://my-call-center/site/login' -o test.py to open an chrome GUI for testing call.

I have run PulseAudio with profile as below (cat ~/.config/pulse/default.pa):

.include /etc/pulse/default.pa

load-module module-pipe-sink file=/tmp/fifo.out
load-module module-pipe-source channels=2 file=/tmp/fifo.in

set-default-source fifo_input
set-default-sink fifo_output

This will create pipe file /tmp/fifo.out and when I call on the website, the chrome will use /tmp/fifo.out as output (as a speaker). I use cat /tmp/fifo.out | tee output.out to got the audio content.

But the output.out file is too large and couldn't be listen. After I removed all the \x00\x00 in this file (with python), I could be able to hear some words in the file but still not well. (I use Audacity APP to import the file to listen.)

So, I want to know how can I correctly output the audio from PulseAudio into a wave format file.

pulseaudio version 15.99.1

By lots of trials. From the practical law and some experience, I properly handled this problem.

Just need read the fifo.out file as solid frequence and read solid chunk of data.

The script are below:

import os
import time

stream_out_file = '/tmp/fifo.out'
output_file = 'py_t.wav'

if not output_file.endswith('.wav'):
    output_file = output_file + '.wav'

# 经计算 16bit 2ch 44.1kHz 下，每秒读取数据 176400B
# 所以，200ms 有数据 35280B，20ms 有 3528B
# 系统 buf 大小 64k，为 65536B，约 18.57ms (无用)
stream = os.open(stream_out_file, os.O_RDONLY )
with open(output_file, 'wb') as wav_fd:
    start = time.time_ns()                   # start time
    _next = time.time_ns() + 20 * 1000*1000 # next time is after 20ms
    while True:
        if time.time_ns() > _next:   # 每 20ms 读取一次
            _next += 20 * 1000*1000 # next time
            s = os.read(stream, 3528)
            print("read length", len(s))
            wav_fd.write(s)
        else: # 睡 2ms
            time.sleep(0.002)

The numbers above are calculated from the pulseaudio configure. My configures are as below:

.include /etc/pulse/default.pa

load-module module-pipe-sink channels=2 rate=44100 format=s16le file=/tmp/fifo.out
load-module module-pipe-source channels=2  file=/tmp/fifo.in

set-default-source fifo_input
set-default-sink fifo_output

As to why, I can't explain it.