respeaker_python_library
respeaker_python_library copied to clipboard
respeaker
snowboy wake word detection is not working with respeaker 2 mic pi hat.
i've been trying to make respeaker detect my wakeword for 2 days .still cant make it out .i thing the problem is with 16bit mono recording all i can hear is a white noise .but stereo recording is working fine no issue at all.tried pyaudio ,arecord . all of them have same problem while mono recording .
can some one help me fix this problem. i can provide any info u guys want
Have you tried to use audacity to find out if it's the case? Maybe one microphone is broken
Which ALSA device do you select? The default will support mono recording
no its not working . do i need to make any changes to .asounrc file
pcm.!default { type hw card 1 format S16_LE }
ctl.!default { type hw card 1 }
``
these are my device info
============================== Default recording device number: 9 Default playback device number: 9
Device ID: 0 Device name: bcm2835 ALSA: - (hw:0,0) Host name: ALSA Recording channels: 0 Playback channels: 2 Low Recording Latency: -1.000000 Low Playback Latency: 0.005805 High Recording Latency: -1.000000 High Playback Latency: 0.034830 Supported Rates: 8000 9600 11025 12000 15000 16000 22050 24000 32000 44100 48000
Device ID: 1 Device name: bcm2835 ALSA: IEC958/HDMI (hw:0,1) Host name: ALSA Recording channels: 0 Playback channels: 2 Low Recording Latency: -1.000000 Low Playback Latency: 0.005805 High Recording Latency: -1.000000 High Playback Latency: 0.034830 Supported Rates: 44100 48000
Device ID: 2 Device name: seeed-2mic-voicecard: - (hw:1,0) Host name: ALSA Recording channels: 2 Playback channels: 2 Low Recording Latency: 0.005805 Low Playback Latency: 0.008707 High Recording Latency: 0.034830 High Playback Latency: 0.034830 Supported Rates: 8000 11025 16000 22050 32000 44100 48000
Device ID: 3 Device name: sysdefault Host name: ALSA Recording channels: 0 Playback channels: 128 Low Recording Latency: -1.000000 Low Playback Latency: 0.005805 High Recording Latency: -1.000000 High Playback Latency: 0.034830 Supported Rates: 8000 9600 11025 12000 15000 16000 22050 24000 32000 44100 48000 88200 176400
Device ID: 4 Device name: playback Host name: ALSA Recording channels: 0 Playback channels: 128 Low Recording Latency: -1.000000 Low Playback Latency: 0.125000 High Recording Latency: -1.000000 High Playback Latency: 0.125000 Supported Rates: 8000 9600 12000 15000 16000 24000 32000 44100 48000 88200 96000 176400 192000
Device ID: 5 Device name: capture Host name: ALSA Recording channels: 128 Playback channels: 0 Low Recording Latency: 0.125000 Low Playback Latency: -1.000000 High Recording Latency: 0.125000 High Playback Latency: -1.000000 Supported Rates:
Device ID: 6 Device name: dmixed Host name: ALSA Recording channels: 0 Playback channels: 2 Low Recording Latency: -1.000000 Low Playback Latency: 0.125000 High Recording Latency: -1.000000 High Playback Latency: 0.125000 Supported Rates: 48000
Device ID: 7 Device name: array Host name: ALSA Recording channels: 2 Playback channels: 0 Low Recording Latency: 0.125000 Low Playback Latency: -1.000000 High Recording Latency: 0.125000 High Playback Latency: -1.000000 Supported Rates:
Device ID: 8 Device name: dmix Host name: ALSA Recording channels: 0 Playback channels: 2 Low Recording Latency: -1.000000 Low Playback Latency: 0.021333 High Recording Latency: -1.000000 High Playback Latency: 0.021333 Supported Rates: 48000
Device ID: 9 Device name: default Host name: ALSA Recording channels: 2 Playback channels: 2 Low Recording Latency: 0.005805 Low Playback Latency: 0.008707 High Recording Latency: 0.034830 High Playback Latency: 0.034830 Supported Rates: 8000 11025 16000 22050 32000 44100 48000
Selected recording device: 9 - default Selected playback device: 9 - default Supported Rates: 8000 11025 16000 22050 32000 44100 48000
Available mixers:
Available recording sources:
Available playback volumes: 0 - PCM:0
Recording volume is emulated Playback volume is emulated
default recording device can record in stereo with no issue . but in the case of mono something is not right ..
this is the screen of audacity when i'm recording in stereo .take a look at audio meter
this one when recording mono
can some one please help me fix this issue .i did a fresh install of raspbian stretch today .still the problem exist
Are the volumes of the two microphone the same?
You can use alsamixer -c 1 to set the volumes.
If you want to to restore the default volumes, try sudo alsactl --file=/etc/voicecard/wm8960_asound.state restore
By the way, you can remove your alsa configuration file ~/.asound.rc and just use the default alsa configuration file /etc/asound.conf from the driver
@xiongyihui i Tried all of the steps u've mentioned . still the problem exist these are my alsa-infos alsa - info : http://www.alsa-project.org/db/?f=bfd07e437056c0a9560290340fa9b4c0a5ade7e1
as you guys can see audio recorded from left input mic is just too noisy .but audio from right one working perfectly . i can findout this by playing tracks individually .but when i play them together there is no issue audio was perfect .
And i noticed that when i'm recording mono left input mic is activated . is there anyway i can use my right mic when recording mono ?.
like this one mentioned in this thread https://unix.stackexchange.com/questions/254219/alsa-capturing-mono-microphone-as-right-channel-of-stereo
my mic hat was just 2 moths old i'm from india and i bought it from indian reseller https://www.fabtolab.com/respeaker-hat my mic hat sometimes automatically turn on its led lights .i need to unplug and replug my rp3 to the power . sometime i need to plugin rp3 adapter to different sockets to make the lights turn off . does this relates to my issue ? is there any firmware updates available for me @xiongyihui any help?
def get_callback(self):
def callback(in_data, frame_count, time_info, status):
sound = AudioSegment(in_data,sample_width=2, channels=1, frame_rate=16000)
self.wavefile.writeframes(sound.raw_data)
return in_data, pyaudio.paContinue
return callback
by this pyaudio callback function i can convert stereo audio to mono in realtime using pydub .but the problem is only 32bit pcm is working . but when i change wavefile.setsampwidth(4) wavefile sampwidth to 2 ie 16bit pcm it becomes completely laggy . any way to fix ?
@arjunan47 Did you want to save audio data to .wav file? You can use wave module, See https://docs.python.org/2/library/wave.html
import wave
wav = wave.open('audio.wav', 'wb')
# ...
PS. As the left channel is broken, before it is fixed, you can recording stereo audio data and select the right channel to use. To get right channel from stereo:
import audioop
stereo = '' # stereo audio data
width = 2
right = audioop.tomono(stereo, width, 0, 1)
@xiongyihui bro is it possible to record audio in realtime using buffers/chunks and extract mono from it?.may be I could use it for wake word detection using snowboy
i don't what i did wrong , one day it starts to work .nextday it stops .
``
import pyaudio
import wave
import time
import audioop
import snowboydetect
import threading
import Queue
detector = snowboydetect.SnowboyDetect(resource_filename = '/home
/pi/xxxxxcommon.res'.encode(),model_str ='/home/xxxxx.umdl'.encode())
class SnowboyRec(threading.Thread):
def __init__(self,channels = 2,
rate = 16000, frames_per_buffer = 1024):
threading.Thread.__init__(self)
self.channels = channels
self.rate = rate
self.frames_per_buffer = frames_per_buffer
self._pa = pyaudio.PyAudio()
self._stream = None
def start_recording(self):
detector = snowboydetect.SnowboyDetect(resource_filename = 'xxxx'.encode(),model_str ='
xxxxxxxx'.encode())
detector.SetAudioGain(9)
detector.ApplyFrontend(False)
detector.SetSensitivity('0.9'.encode())
print('entered')
self._stream = self._pa.open(format=pyaudio.paInt16,
channels=self.channels,
rate=self.rate,
input_device_index = 2,
input=True,
frames_per_buffer=self.frames_per_buffer,
stream_callback=self.get_callback())
print('start')
self._stream.start_stream()
return self
def get_callback(self):
def callback(in_data, frame_count, time_info, status):
d = audioop.tomono(in_data, 2, 0, 1)
status = detector.RunDetection(d)
print(status)
if status == 1:
print('keyword spotted')
return in_data, pyaudio.paContinue
return callback
if name == 'main': snow = SnowboyRec() snow.start_recording() it works when i speak so loud `