'pyaudio save multiple .WAV file with nonblocking
Updates: Now I found out that we can paste some code in the callback function and ended up more questions:
When will be call and stop the callback functions? when we open and close the stream?
The callback function can return the stream data(audio_data from the code). As we did not call the function, the pyaudio do it internally I believe. How do I get the return stream data from callback?
import pyaudio
import wave
import numpy as np
import npstreams
import time
CHUNK = 1024
FORMAT = pyaudio.paInt16
CHANNELS = 2
RATE = 44100
audio = pyaudio.PyAudio()
channel_1_frames = []
channel_2_frames = []
fulldata = np.array([])
def callback(in_data, frame_count, time_info, flag):
global b,a,fulldata #global variables for filter coefficients and array
audio_data = np.fromstring(in_data, dtype=np.int16)
channel_1 = audio_data[0::CHANNELS]
channel_2 = audio_data[1::CHANNELS]
data1 = channel_1.tostring()
data2 = channel_2.tostring()
channel_1_frames.append(data1)
channel_2_frames.append(data2)
wf1 = wave.open('Channel_1.wav', 'wb')
wf2 = wave.open('Channel_2.wav', 'wb')
wf1.setnchannels(1)
wf2.setnchannels(1)
wf1.setsampwidth(audio.get_sample_size(FORMAT))
wf2.setsampwidth(audio.get_sample_size(FORMAT))
wf1.setframerate(RATE)
wf2.setframerate(RATE)
wf1.writeframes(b''.join(channel_1_frames))
wf2.writeframes(b''.join(channel_2_frames))
wf1.close()
wf2.close()
return (audio_data, pyaudio.paContinue)
stream = audio.open(format=FORMAT,
channels=CHANNELS,
rate=RATE,
input=True,
#frames_per_buffer=CHUNK,
stream_callback=callback)
stream.start_stream()
while stream.is_active():
time.sleep(10)
stream.stop_stream()
stream.close()
audio.terminate()
=============================================
I am trying to record multiple channels into multiple .WAV file. I can do that with stream.read() and numpy array to separate into different array, and save to .WAV file
stream = audio.open(format=FORMAT,
channels=CHANNELS,
rate=RATE,
input=True,
frames_per_buffer=CHUNK)
print("* recording")
channel_1_frames = []
channel_2_frames = []
for i in range(0, int(RATE / CHUNK * RECORD_SECONDS)):
data = stream.read(CHUNK)
# convert string to numpy array
data_array = np.frombuffer(data, dtype='int16')
# select channel
channel_1 = data_array[0::CHANNELS]
channel_2 = data_array[1::CHANNELS]
# convert numpy array to string
data1 = channel_1.tostring()
data2 = channel_2.tostring()
channel_1_frames.append(data1)
channel_2_frames.append(data2)
stream.stop_stream()
stream.close()
audio.terminate()
However, from the module documentation, https://people.csail.mit.edu/hubert/pyaudio/docs/#class-stream, it said stream.read() and stream.write() should not be used for non-blocking.
And I found a good non-blocking pyaudio example from Github: https://gist.github.com/sloria/5693955 Which did not use stream.read().
I am not sure if I can read and turn steam numpy array without stream.read() So is it still possible to export the stream in to different .WAV? and make it non blocking?
Thanks
Solution 1:[1]
As I learn more in coding, I found the answers.
A1: The callback function run and stop with the stream.
######open stream with out starting automatically
audio = pyaudio.PyAudio()
stream = audio.open(format=format,
channels=2,
rate=44100,
input=True,
frames_per_buffer=44100,
stream_callback=self.get_callback(),
start=False)
######start,stop stream
stream.start_stream()
stream.close()
audio.terminate()
A2: To capture data in real time, we can use queue
self.recorded_frames = queue.Queue()
def get_callback(self):
def callback(in_data, frame_count, time_info, status):
self.recorded_frames.put(np.frombuffer(in_data, dtype=np.int16))
return in_data, pyaudio.paContinue
return callback
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
Solution | Source |
---|---|
Solution 1 | Giorgos Xou |