def main():
rospy.init_node( 'beep_publisher' )
# parameters
loop_rate = rospy.get_param( '~loop_rate', 100 )
buffer_size = rospy.get_param( '~buffer_size', 4096 )
use_audio_common = rospy.get_param( '~use_audio_common', False )
topicname_data = '~audio'
topicname_info = '~info'
# import message modules
if use_audio_common:
from audio_common_msgs.msg import AudioData
from audio_stream_msgs.msg import AudioInfo
else:
from audio_stream_msgs.msg import AudioData, AudioInfo
# publishers
publisher_data = rospy.Publisher( topicname_data, AudioData, queue_size=1000 )
publisher_info = rospy.Publisher( topicname_info, AudioInfo, queue_size=10 )
# message
msg_data = AudioData()
msg_info = AudioInfo()
# gen a default buffer
msg_info.channels = 1
msg_info.sampling_rate = 16000
msg_info.width = 2
frequency_f0 = 440 # 基本周波数
frequency_beep = 1.0 # beepの周期
buffer_beep = generateBeep( 0.1*((2**8) ** 2), [ 440, 480, 500, 520, 400, 300, 600, 800 ], frequency_beep, msg_info.sampling_rate, 10.0, 0.5 ).astype(np.uint16).tobytes()
r = rospy.Rate( loop_rate )
index = 0
time_played = 0.0; # sec
time_start = time.time()
while not rospy.is_shutdown():
rospy.loginfo('publish')
if index > len(buffer_beep):
index = 0
msg_info.header.stamp = rospy.Time.now()
if ( len(buffer_beep) - index) < buffer_size:
msg_data.data = buffer_beep[index:]
time_played += 1.0 * ( len(buffer_beep) - index ) / ( msg_info.width * msg_info.sampling_rate )
else:
msg_data.data = buffer_beep[index:index+buffer_size]
time_played += 1.0 * buffer_size / ( msg_info.width * msg_info.sampling_rate )
time_passed = time.time() - time_start
while time_played - time_passed > 0:
time_passed = time.time() - time_start
if rospy.is_shutdown():
return
r.sleep()
publisher_data.publish( msg_data )
publisher_info.publish( msg_info )
index += buffer_size
def audio_cb(self, data):
#: :type data: AudioBuffer
#ab = AudioBuffer()
# This contains (in the 'data' field): int16[] data
rospy.loginfo("Callback received!")
ad = AudioData()
# This is: uint8[] data
data_uint8 = numpy.array(data.data, dtype=numpy.uint8)
ad.data = data_uint8.tolist()
self.pub.publish(ad)
def sample_audio_test(self, in_data, frame_count, time_info, status):
#data_int = np.fromstring(in_data, dtype=np.int16)
#msg_audio = AudioData16()
#msg_audio.data = data_int
msg_audio = AudioData()
msg_audio.data = in_data
self.raw_pub.publish(msg_audio)
return None, pyaudio.paContinue
def tts(self, words):
result = self.client.synthesis(words, 'zh', 1, {
'vol': 5,
})
if isinstance(result, dict):
rospy.logerr(result)
return None
audio_data = AudioData()
audio_data.data = result
return audio_data
def read_from_cache(text):
if not os.path.exists("/tmp/xiaoqiang_tts"):
os.mkdir("/tmp/xiaoqiang_tts")
return None
if os.path.exists(os.path.join("/tmp/xiaoqiang_tts", text.data)):
with open(os.path.join("/tmp/xiaoqiang_tts", text.data),
"rb") as audio_file:
audio_data = AudioData()
audio_data.data = audio_file.read()
return audio_data
return None
def on_audio(self, data, channel):
self.pub_audios[channel].publish(AudioData(data=data))
if channel == self.main_channel:
self.pub_audio.publish(AudioData(data=data))
if self.is_speeching:
if len(self.speech_audio_buffer) == 0:
self.speech_audio_buffer = self.speech_prefetch_buffer
self.speech_audio_buffer += data
else:
self.speech_prefetch_buffer += data
self.speech_prefetch_buffer = self.speech_prefetch_buffer[-self.speech_prefetch_bytes:]
def stream_audio(self):
reccmd = self.arecord_command
self.p = subprocess.Popen(reccmd, stdout=subprocess.PIPE)
while not rospy.is_shutdown():
self.last_data = self.p.stdout.read(self.block_size)
if self.last_data is not None:
ad = AudioData()
ad.data = tuple(np.fromstring(self.last_data, dtype=np.uint8))
self.pub.publish(ad)
self.close()
def run(self):
try:
data = self.stream.read(self.CHUNK)
except IOError as ex:
if ex[1] != pyaudio.paInputOverflowed:
raise
print("ERROR")
return
amplitude = np.fromstring(data, np.uint8)
print type(amplitude)
msg = AudioData()
msg.data = amplitude.tolist()
self.pub.publish(msg)
def read_from_cache(text):
m = hashlib.md5()
m.update(text.data)
audio_filename = m.hexdigest()
if not os.path.exists("xiaoqiang_tts"):
os.mkdir("xiaoqiang_tts")
return None
if os.path.exists(os.path.join("xiaoqiang_tts", audio_filename)) and \
os.path.getsize(os.path.join("xiaoqiang_tts", audio_filename)) > 0:
with open(os.path.join(
"xiaoqiang_tts", audio_filename), "rb") as audio_file:
audio_data = AudioData()
audio_data.data = audio_file.read()
return audio_data
return None
def topic_cb(self, msg):
# Retrieve 1 channel audio data
tmp = list(msg.data)
tmp = tmp[0::len(msg.channelMap)]
# Convert audio format: Int -> Buffer(Uint8)
dataBuff = ""
for i in range(0, len(tmp)):
if tmp[i] < 0:
tmp[i] = tmp[i] + 65536
dataBuff = dataBuff + chr(tmp[i] % 256)
dataBuff = dataBuff + chr((tmp[i] - (tmp[i] % 256)) / 256)
# Publish audio topic
audio_data = AudioData()
audio_data.data = dataBuff
self.audiodata_pub.publish(audio_data)
def main():
rospy.init_node('AudioCapturer', anonymous=True)
publisher = rospy.Publisher("rawAudioChunk", AudioData, queue_size=20)
p = pyaudio.PyAudio()
stream = p.open(
input_device_index=None, # Default device
format=FORMAT,
channels=CHANNELS,
rate=RATE,
input=True,
frames_per_buffer=CHUNK_SIZE)
print("*Recording*")
# Loop while node is not closed and audio is working. Note that
# here is "not needed" something like `spin()` or `loop_sleep()`
# because (presumably) PyAudio manages the blocked times waiting for
# IO and puts the process/thread to sleep, also there are no callbacks
# by ROS.
while stream.is_active() and not rospy.is_shutdown():
try:
in_data = stream.read(CHUNK_SIZE, exception_on_overflow=False)
msg = in_data
publisher.publish(AudioData(data=msg))
except IOError as e:
print("I/O error({0}): {1}".format(e.errno, e.strerror))
# break
if not stream.is_active():
print("Stream was not active.")
stream.stop_stream()
stream.close()
p.terminate()
def onInitialize(self):
#rospy.loginfo(self._properties.getProperty("type_name") + " version " + self._properties.getProperty("version"))
rospy.loginfo("Copyright (C) 2010-2011 Yosuke Matsusaka")
rospy.loginfo("Copyright (C) 2017 Isao Hara")
rospy.loginfo("Copyright (C) 2020 Isao Hara")
self._prevtime = time.clock()
# configuration parameters
self._samplerate = [16000,]
self.bindParameter("rate", self._samplerate, 16000)
self._character = ["male",]
self.bindParameter("character", self._character, "male")
self._sampling_rate = [0,]
self.bindParameter("sampling_rate", self._sampling_rate, 0)
# create inport
self._tts_sub = rospy.Subscriber('/tts/data', String, self.onData)
# create outport for wave data
self._audio_out = rospy.Publisher('/audio_play/audio', AudioData, queue_size=1)
self._outdata = AudioData()
self._status = "ready"
self._is_active = False
return True
def publish(self):
# Input Data
input_data = self.audio_data_buffer[:self.length_queue_popup]
self.audio_data_buffer = self.audio_data_buffer[self.length_queue_popup:]
try:
is_speech = self._vad.is_speech(
input_data, self._audio_info.sample_rate)
except Exception as e:
rospy.logerr('Got an error while processing. {}'.format(e))
return
self._pub_is_speech.publish(Bool(is_speech))
if self._current_speaking == True and is_speech == True:
self._speech_audio_buffer += input_data
elif self._current_speaking == False and is_speech == True:
self._speech_audio_buffer = input_data
self._current_speaking = True
elif self._current_speaking == True and is_speech == False:
self._speech_audio_buffer = self._speech_audio_buffer + input_data
speech_duration = (len(self._speech_audio_buffer) /
2.0) / self._audio_info.sample_rate
if speech_duration > self._minimum_duration:
self._pub_speech_audio.publish(
AudioData(self._speech_audio_buffer))
else:
rospy.logwarn(
'speech duration: {} dropped'.format(speech_duration))
self._current_speaking = False
self._speech_audio_buffer = None
rospy.logwarn('Cleared speech_audio_buffer.')
def on_timer(self, event):
stamp = event.current_real or rospy.Time.now()
is_voice = self.respeaker.is_voice()
doa_rad = math.radians(self.respeaker.direction - 180.0)
doa_rad = angles.shortest_angular_distance(
doa_rad, math.radians(self.doa_yaw_offset))
doa = math.degrees(doa_rad)
# vad
if is_voice != self.prev_is_voice:
self.pub_vad.publish(Bool(data=is_voice))
self.prev_is_voice = is_voice
# doa
if doa != self.prev_doa:
self.pub_doa_raw.publish(Int32(data=doa))
self.prev_doa = doa
msg = PoseStamped()
msg.header.frame_id = self.sensor_frame_id
msg.header.stamp = stamp
ori = T.quaternion_from_euler(math.radians(doa), 0, 0)
msg.pose.position.x = self.doa_xy_offset * np.cos(doa_rad)
msg.pose.position.y = self.doa_xy_offset * np.sin(doa_rad)
msg.pose.orientation.w = ori[0]
msg.pose.orientation.x = ori[1]
msg.pose.orientation.y = ori[2]
msg.pose.orientation.z = ori[3]
self.pub_doa.publish(msg)
if self.asr_engine != "silent":
#actual cutting of the "speech" only audio for sending to asr service
# get timestamp of "last time mic array detected voice"
if is_voice and not self.pepper_is_speaking:
self.speech_stopped = stamp
#pepper is speaking: we go to cut (correct)
#pepper is not speaking:
#we are past the grace period in speaking:
#second statement fails, we go to cutting (correct)
#we are not past the grace period in speaking:
#second statement is true, we don't cut off audio and go on with is_speaking = True (correct)
if not self.pepper_is_speaking and (stamp - self.speech_stopped <
rospy.Duration(
self.speech_continuation)):
self.is_speaking = True #grace period "speech_continuation where we just continue 'is speaking'"
elif self.is_speaking: #otherwise there has been no speech for too long, and we were obviously speaking, so we cut the audio here
buf = self.speech_audio_buffer
self.speech_audio_buffer = str()
self.is_speaking = False
duration = 8.0 * len(buf) * self.respeaker_audio.bitwidth
duration = duration / self.respeaker_audio.rate / self.respeaker_audio.bitdepth
rospy.loginfo("Speech detected for %.3f seconds" % duration)
if self.speech_min_duration <= duration < self.speech_max_duration:
self.pub_speech_audio.publish(AudioData(data=buf))
def on_audio(self, data):
self.pub_audio.publish(AudioData(data=data))
if self.is_speeching:
if len(self.speech_audio_buffer) == 0:
self.speech_audio_buffer = self.speech_prefetch_buffer
self.speech_audio_buffer += data
else:
self.speech_prefetch_buffer += data
self.speech_prefetch_buffer = self.speech_prefetch_buffer[-self.speech_prefetch_bytes:]
def read_from_cache(text):
global locale
m = hashlib.md5()
m.update(text.data)
audio_filename = m.hexdigest()
if not os.path.exists("xiaoqiang_tts"):
os.mkdir("xiaoqiang_tts")
return None
# 兼容以前版本
if os.path.exists(os.path.join("xiaoqiang_tts", audio_filename)) and \
os.path.getsize(os.path.join("xiaoqiang_tts", audio_filename)) > 0 and locale == "zh-cn":
with open(os.path.join("xiaoqiang_tts", audio_filename),
"rb") as audio_file:
audio_data = AudioData()
audio_data.data = audio_file.read()
return audio_data
if not os.path.exists(os.path.join("xiaoqiang_tts", locale)):
os.mkdir(os.path.join("xiaoqiang_tts", locale))
return None
# 从对应locale路径读取
if os.path.exists(os.path.join("xiaoqiang_tts", locale, audio_filename)) and \
os.path.getsize(os.path.join("xiaoqiang_tts", locale, audio_filename)) > 0:
with open(os.path.join("xiaoqiang_tts", locale, audio_filename),
"rb") as audio_file:
audio_data = AudioData()
audio_data.data = audio_file.read()
return audio_data
return None
def on_audio(self, data):
speech_data = None
for i, ch in enumerate(self.channels):
chan_data = data[:, ch].tostring()
if not speech_data:speech_data = chan_data
self.pub_audio[i].publish(AudioData(data=chan_data))
if self.is_speeching:
if len(self.speech_audio_buffer) == 0:
self.speech_audio_buffer = self.speech_prefetch_buffer
self.speech_audio_buffer += speech_data
else:
self.speech_prefetch_buffer += speech_data
self.speech_prefetch_buffer = self.speech_prefetch_buffer[-self.speech_prefetch_bytes:]
def on_timer(self, event):
stamp = event.current_real or rospy.Time.now()
is_voice = self.respeaker.is_voice()
doa_rad = math.radians(self.respeaker.direction - 180.0)
doa_rad = angles.shortest_angular_distance(
doa_rad, math.radians(self.doa_yaw_offset))
doa = math.degrees(doa_rad)
# vad
if is_voice != self.prev_is_voice:
self.pub_vad.publish(Bool(data=is_voice))
self.prev_is_voice = is_voice
# doa
if doa != self.prev_doa:
self.pub_doa_raw.publish(Int32(data=doa))
self.prev_doa = doa
msg = PoseStamped()
msg.header.frame_id = self.sensor_frame_id
msg.header.stamp = stamp
ori = T.quaternion_from_euler(math.radians(doa), 0, 0)
msg.pose.position.x = self.doa_xy_offset * np.cos(doa_rad)
msg.pose.position.y = self.doa_xy_offset * np.sin(doa_rad)
msg.pose.orientation.w = ori[0]
msg.pose.orientation.x = ori[1]
msg.pose.orientation.y = ori[2]
msg.pose.orientation.z = ori[3]
self.pub_doa.publish(msg)
# speech audio
if is_voice:
self.speech_stopped = stamp
if stamp - self.speech_stopped < rospy.Duration(
self.speech_continuation):
self.is_speeching = True
elif self.is_speeching:
buf = self.speech_audio_buffer
self.speech_audio_buffer = str()
self.is_speeching = False
duration = 8.0 * len(buf) * self.respeaker_audio.bitwidth
duration = duration / self.respeaker_audio.rate / self.respeaker_audio.bitdepth
rospy.loginfo("Speech detected for %.3f seconds" % duration)
if self.speech_min_duration <= duration < self.speech_max_duration:
self.pub_speech_audio.publish(AudioData(data=buf))
def onInitialize(self):
rospy.loginfo("GoogleTextToSpeechRos version " + __version__)
rospy.loginfo("Copyright (C) 2019 Isao Hara")
#
# create inport for audio stream
self._inport = rospy.Subscriber('/tts/data', String, self.onData)
#
# create outport for result
self._outdata = AudioData()
self._outport = rospy.Publisher('/audio_play/audio',
AudioData,
queue_size=10)
#
#
self.show_copyrights()
return True
def send_audio(self):
#print bytearray(buff)
#print len(buff)
#print len(buff)/200
with contextlib.closing(wave.open("test.wav", 'wb')) as wf:
wf.setnchannels(1)
wf.setsampwidth(2)
wf.setframerate(16000)
wf.writeframes(bytearray(self.tmp_buff))
tmp = []
with contextlib.closing(open("test.wav", 'rb')) as wf:
ba = wf.read()
for c in ba:
#print c
value = struct.unpack('B', c)[0]
#print value
tmp.append(value)
self.pub_audio.publish(AudioData(tmp))
self.tmp_buff = None
def main():
rospy.init_node('colored_noise_publisher')
# parameters
loop_rate = int(rospy.get_param('~loop_rate', 100))
buffer_size = int(rospy.get_param('~buffer_size', 4096))
use_audio_common = bool(rospy.get_param('~use_audio_common', False))
beta = float(rospy.get_param('~exponent', 1))
level = float(rospy.get_param('~level', 0.5))
# import message modules
if use_audio_common:
from audio_common_msgs.msg import AudioData
from audio_stream_msgs.msg import AudioInfo
else:
from audio_stream_msgs.msg import AudioData, AudioInfo
# publishers
publisher_data = rospy.Publisher('~audio', AudioData, queue_size=1000)
publisher_info = rospy.Publisher('~info', AudioInfo, queue_size=10)
# gen a default buffer
msg_info = AudioInfo()
msg_data = AudioData()
msg_info.channels = 1
msg_info.sampling_rate = 16000
msg_info.width = 2
A = level * ((2**8)**2)
buffer_wave = generateColoredNoise(beta, A, msg_info.sampling_rate,
30).astype(np.uint16).tobytes()
r = rospy.Rate(loop_rate)
index = 0
time_played = 0.0 # sec
time_start = time.time()
while not rospy.is_shutdown():
rospy.loginfo('publish')
# update msg_info header
msg_info.header.stamp = rospy.Time.now()
# fill msg_data.data
if (len(buffer_wave) - index) < buffer_size:
msg_data.data = buffer_wave[index:] + buffer_wave[:index +
buffer_size -
len(buffer_wave)]
time_played += 1.0 * (len(buffer_wave) - index) / (
msg_info.width * msg_info.sampling_rate)
index = index + buffer_size - len(buffer_wave)
else:
msg_data.data = buffer_wave[index:index + buffer_size]
time_played += 1.0 * buffer_size / (msg_info.width *
msg_info.sampling_rate)
index = index + buffer_size
time_passed = time.time() - time_start
while time_played > time_passed:
time_passed = time.time() - time_start
if rospy.is_shutdown():
return
r.sleep()
# publish them
publisher_data.publish(msg_data)
publisher_info.publish(msg_info)
def on_raw_audio(self, data, mic):
pub = self.raw_audio_pubs[mic]
pub.publish(AudioData(data=data))
def __init__(self):
self.sub_audio = rospy.Subscriber('/audio', AudioData, self.audioCB)
self.audio = AudioData()
import time
from pyhelper_fns.vis_utils import MyAnimationMulti
import numpy as np
import time
import pickle
import argparse
def gray_rgb(image):
return np.concatenate([np.expand_dims(image, 2), np.expand_dims(image, 2), np.expand_dims(image, 2)], axis = 2)
vis_tool = MyAnimationMulti(None, numPlots=6, isIm = np.ones(6))
if __name__ == '__main__':
rospy.init_node('traj_replay_audio', anonymous = True)
pub = rospy.Publisher('replay_audio/audio', AudioData, queue_size = 10)
msg = AudioData()
parser = argparse.ArgumentParser()
parser.add_argument('path', type=str,
help='path to pickle file')
args = parser.parse_args()
FILE_NAME = args.path
with open(FILE_NAME, "rb") as handle:
data_dict = pickle.load(handle)
states = data_dict["state_list"]
actions = data_dict["action_list"]
images = data_dict["image_list"]
audio = data_dict["audio_list"]
print("len of states{} len of actions{} len of images{}".format(len(states), len(actions), len(images)))
def main():
rospy.init_node( 'audio_stream_publisher' )
# parameters
loop_rate = rospy.get_param( '~loop_rate', 100 )
is_loop = rospy.get_param( '~is_loop', False )
filename = rospy.get_param( '~filename' )
buffer_size = rospy.get_param( '~buffer_size', 4096 )
max_precede = rospy.get_param( '~max_precede', 10.0 )
use_audio_common = rospy.get_param( '~use_audio_common', False )
topicname_data = '~audio'
topicname_info = '~info'
# import message modules
if use_audio_common:
from audio_common_msgs.msg import AudioData
from audio_stream_msgs.msg import AudioInfo
else:
from audio_stream_msgs.msg import AudioData, AudioInfo
# publishers
publisher_data = rospy.Publisher( topicname_data, AudioData, queue_size=1 )
publisher_info = rospy.Publisher( topicname_info, AudioInfo, queue_size=1 )
# message
msg_data = AudioData()
msg_info = AudioInfo()
# open wavefile
wavefile = wave.open( filename, 'r' )
wave_channels = wavefile.getnchannels()
wave_width = wavefile.getsampwidth()
wave_framerate = wavefile.getframerate()
wave_framenum = wavefile.getnframes()
wave_data = wavefile.readframes( wave_framenum )
infostr = (\
'== Wave file info ==\n' +\
'channels: {}\n' +\
'sample size [byte]: {}\n' +\
'sampling rate [hz]: {}\n' +\
'number of frames: {} ( {} secs )').format( wave_channels, wave_width, wave_framerate, wave_framenum, wave_framenum*1.0/wave_framerate )
rospy.loginfo( infostr )
if wave_channels != 1 or not ( wave_width == 1 or wave_width == 2):
rospy.logerr( 'Currently 8bit or 16bit monaural wave files are supported.' )
sys.exit(1)
if buffer_size % wave_width > 0:
rospy.logerr ( 'buffer_size must be multiple of wave_width' )
sys.exit(1)
msg_info.channels = wave_channels
msg_info.sampling_rate = wave_framerate
msg_info.width = wave_width
msg_info.header.stamp = rospy.Time.now()
publisher_info.publish( msg_info )
r = rospy.Rate( loop_rate )
if is_loop:
time_played = 0.0; # sec
time_start = time.time()
index = 0
while not rospy.is_shutdown():
# return to first if index is out of wave_data
if index > wave_framenum:
index = 0
# preparing msg data
msg_info.header.stamp = rospy.Time.now()
if ( wave_framenum - index ) < buffer_size:
msg_data.data = wave_data[index:]
time_played += 1.0 * ( wave_framenum - index ) / ( wave_width * wave_framerate )
else:
msg_data.data = wave_data[index:index+buffer_size]
time_played += 1.0 * buffer_size / ( wave_width * wave_framerate )
# sleep if played time is ahead of passed time
time_passed = time.time() - time_start
while time_played - time_passed > max_precede:
time_passed = time.time() - time_start
if rospy.is_shutdown():
return
r.sleep()
# publishing data
publisher_data.publish( msg_data )
publisher_info.publish( msg_info )
# debug print
rospy.logdebug( 'time played: {}, time passed {}'.format( time_played, time_passed ) )
# increase index
index += buffer_size
else:
time_played = 0.0; # sec
time_start = time.time()
index = 0
while not rospy.is_shutdown():
# return to first if index is out of wave_data
if index > wave_framenum:
break
# preparing msg data
msg_info.header.stamp = rospy.Time.now()
if ( wave_framenum - index ) < buffer_size:
msg_data.data = wave_data[index:]
time_played += 1.0 * ( wave_framenum - index ) / ( wave_width * wave_framerate )
else:
msg_data.data = wave_data[index:index+buffer_size]
time_played += 1.0 * buffer_size / ( wave_width * wave_framerate )
# sleep if played time is ahead of passed time
time_passed = time.time() - time_start
while time_played - time_passed > max_precede:
time_passed = time.time() - time_start
if rospy.is_shutdown():
return
r.sleep()
# publishing data
publisher_data.publish( msg_data )
publisher_info.publish( msg_info )
# debug print
rospy.logdebug( 'time played: {}, time passed {}'.format( time_played, time_passed ) )
# increase index
index += buffer_size
def publishAudio(self):
if self.chunk_count > self.MIN_CHUNKS_AUDIO_LENGTH:
self.publisher.publish(AudioData(self.voiced_frames))
self.discardAudio()
def on_timer(self, event):
stamp = event.current_real or rospy.Time.now()
is_voice = self.respeaker.is_voice()
doa_rad = math.radians(self.respeaker.direction - 180.0)
doa_rad = angles.shortest_angular_distance(
doa_rad, math.radians(self.doa_yaw_offset))
doa = math.degrees(doa_rad)
if self.is_active:
self.wait_command_count += 1
if self.wait_command_count > self.wait_command_thres:
self.is_active = False
self.respeaker.set_led_off()
if self.is_waiting_response:
self.wait_response_count += 1
if (self.wait_response_count > self.wait_response_thres):
self.is_waiting_response = False
self.respeaker.set_led_off()
# vad
# if is_voice != self.prev_is_voice:
# self.pub_vad.publish(Bool(data=is_voice))
# self.prev_is_voice = is_voice
# doa
if doa != self.prev_doa:
self.pub_doa_raw.publish(Int32(data=doa))
self.prev_doa = doa
# msg = PoseStamped()
# msg.header.frame_id = self.sensor_frame_id
# msg.header.stamp = stamp
# ori = T.quaternion_from_euler(math.radians(doa), 0, 0)
# msg.pose.position.x = self.doa_xy_offset * np.cos(doa_rad)
# msg.pose.position.y = self.doa_xy_offset * np.sin(doa_rad)
# msg.pose.orientation.w = ori[0]
# msg.pose.orientation.x = ori[1]
# msg.pose.orientation.y = ori[2]
# msg.pose.orientation.z = ori[3]
# self.pub_doa.publish(msg)
# speech audio
if is_voice:
self.speech_stopped = stamp
if stamp - self.speech_stopped < rospy.Duration(self.speech_continuation):
self.is_speeching = True
elif self.is_speeching:
buf = self.speech_audio_buffer
self.speech_audio_buffer = str()
self.is_speeching = False
duration = 8.0 * len(buf) * self.respeaker_audio.bitwidth
duration = duration / self.respeaker_audio.rate / self.respeaker_audio.bitdepth
rospy.loginfo("Speech detected for %.3f seconds" % duration)
status = self.detector.RunDetection(buf)
if (status > 0):
print("found kws")
self.is_active = True
self.wait_command_count = 0
self.respeaker.set_led_listen()
return
else:
print("not kws")
# self.respeaker.set_led_trace()
if (self.is_active):
self.is_active = False
if self.speech_min_duration <= duration < self.speech_max_duration:
self.respeaker.set_led_think()
self.is_waiting_response = True
self.wait_response_count = 0
self.pub_speech_audio.publish(AudioData(data=buf))
def subscribe(self):
pass
def unsubscribe(self):
pass
if __name__ == '__main__':
rospy.init_node('audio_capture_microphone')
s = SPH0645Audio()
r = rospy.Rate(float(s.rate) / s.frame_per_buffer)
signal.signal(signal.SIGINT, s.kill)
while s.stream.is_active():
if s._connection_status is not True:
# If not subscribed, only read stream to avoid 'IOError: [Errno -9981] Input overflowed'
# https://stackoverflow.com/questions/10733903/pyaudio-input-overflowed
in_data = np.frombuffer(s.stream.read(1024, exception_on_overflow=False), np.int32)
r.sleep()
continue
# Input data to 16 bit
in_data = np.frombuffer(s.stream.read(1024), np.int32)
in_data = in_data >> 14 # This 18bit integer is raw data from microphone
int16_data = (in_data >> 2).astype(np.int16)
# Retreive 1 channel data
chunk_per_channel = len(int16_data) / s.channels
channel_data = int16_data[s.channel::s.channels]
# Publish audio topic
s.pub_audio.publish(AudioData(data=channel_data.tostring()))
r.sleep()
def on_audio(self, data):
self.pub_audio.publish(AudioData(data=data))
#
# Author: Randoms
#
import rospy
from audio_common_msgs.msg import AudioData
from engines.baidu_tts import BaiduTTS as bd_client
from engines.xunfei_tts import XunfeiTTS as xf_client
from std_msgs.msg import String, Bool
import time
import struct
import numpy
import matplotlib.pyplot as plt
from scipy import signal
AUDIO_CACHE = AudioData()
CURRENT_AUDIO = AudioData()
NEW_AUDIO_FLAG = False
MIN_VOLUM = 2000
AUDIO_STATUS = False
b, a = signal.butter(3, 0.10, 'high')
def is_end(audio_data):
last_two_seconds = audio_data.data[2 * -16000 * 2:]
if len(last_two_seconds) < 2 * 16000 * 2:
return False
last_two_seconds = [float(struct.unpack("<h", "".join(
last_two_seconds[2*x:2*x + 2]))[0]) for x in range(0, 2 * 16000)]
sf = signal.filtfilt(b, a, last_two_seconds)
break
sounds = sounds[0]
positions = positions[0]
rate = rospy.Rate(MSG_FREQ)
sound = np.int16(sounds.mean(0) * 32768)
delete_all = Marker()
delete_all.action = Marker.DELETEALL
delete_all.ns = 'sound_sources'
delete_all = MarkerArray(markers=[delete_all])
for t in xrange((sound.shape[0]+frames_per_msg-1)/frames_per_msg):
begin = t*frames_per_msg
end = min((t+1)*frames_per_msg, sound.shape[0])
audio_pub.publish(AudioData(data = sound[begin:end].tostring()))
markers = MarkerArray()
for i, (snd, pos) in enumerate(zip(sounds, positions)):
scale = np.abs(snd[begin:end]).max()
if scale == 0.:
continue
marker = Marker()
marker.header.stamp = rospy.Time.now()
marker.header.frame_id = 'base_footprint'
marker.type = Marker.SPHERE
marker.ns = 'sound_sources'
marker.id = i
marker.pose.position.x = (pos[0] - MAP_SIZE/2. + .5) * resolution
marker.pose.position.y = (pos[1] - MAP_SIZE/2. + .5) * resolution
marker.pose.orientation.w = 1.
def callback(self, data):
rospy.loginfo("Callback received!")
ad = AudioData()
data_uint8 = numpy.array(data.data, dtype=numpy.uint8)
ad.data = data_uint8.tolist()
self.pub.publish(ad)
