Recording audio stream from ROS – data type error

[匿名]

Hello,

I am new to Hark and I am trying to record the audio captured with a HSR robot’s built-in head microphone. The following are the parameters of the stream generated by the robot:

PARAMETERS
 * /audio/audio_capture/bitrate: 128
 * /audio/audio_capture/channels: 1
 * /audio/audio_capture/device: 
 * /audio/audio_capture/format: mp3
 * /audio/audio_capture/sample_rate: 16000
 * /rosdistro: kinetic
 * /rosversion: 1.12.14

NODES
  /audio/
    audio_capture (audio_capture/audio_capture)

I am trying to record the streamed audio with the AudioStreamFromRos node (Attached you find a screenshot of my Hark network). However, when I execute the network I get the following error message:

[ERROR] [1565594592.746777286]: Client [/MY_HARK_MASTER_NODE] wants topic /audio/audio to have datatype/md5sum [hark_msgs/HarkWave/24c5654436a3ff03c563377fdbcc56a1], but our version has [audio_common_msgs/AudioData/f43a8e1b362b75baa741461b46adc7e0]. Dropping connection.

How can I configure the node to make the data type compatible?

[匿名]

Hi Jm,

Can you please attach the actual network file that you used.

Cheers,
Earl

[匿名]

Hi,

here is the network.

[匿名]

Here. There was an upload error in the previous attempt.

[匿名]

You need to use hark_msgs/HarkWave in your workspace.
In your case, audio_common_msgs/AudioData seems to store mp3 data into uint8[] array, so you will first need to expand it to raw PCM data.

Second, the data structure of hark_msgs/HarkWave is as follows.

user@ubuntu:~$ rosmsg show hark_msgs/HarkWave
std_msgs/Header header
  uint32 seq
  time stamp
  string frame_id
int32 count
int32 nch
int32 length
int32 data_bytes
hark_msgs/HarkWaveVal[] src
  float32[] wavedata

wavedata is a raw PCM data array. Since HARK is not aware of the number of bits, it simply casts an integer value to a floating point type. In other words, 123 is 123.0f .

data_bytes is the data size. In other words, it is the size of float (4 bytes) multiplied by the size of wavedata .

length is the number of samples per frame handled by HARK. The initial value of HARK is 512 . Since HARK processed frame by frame, in other words, the size of wavedata must be nch times length=512 .

nch is the number of channels. Your device seems to be 1ch, so it should be 1 . For microphone array data, a larger number will be stored.

count is the frame count. Since HARK processing frame by frame, it is necessary to know what frame the data is. In other words, it is incremented as the frame advances. The first frame number is 0 . Not 1 .

There is a final note. In order to prevent problems in FFT/IFFT processing, etc., the frames processing by HARK are subject to sample overlap processing.
The following image may help you understand.

Best regards,
m.takigahira

[匿名]

I am using the package ros-kinetic-audio-capture to stream the audio from the built-in microphone of the robot.

With this package it is possible to stream audio data in wave format:

PARAMETERS
 * /audio_capture/channels: 1
 * /audio_capture/depth: 16
 * /audio_capture/device: plughw:1,0
 * /audio_capture/format: wave
 * /audio_capture/sample_rate: 16000
 * /rosdistro: kinetic
 * /rosversion: 1.12.14

NODES
  /
    audio_capture (audio_capture/audio_capture)

The data structure of the data stream audio_common_msgs/AudioData is as follows:

user@laptop:~$ rosmsg show audio_common_msgs/AudioData 
uint8[] data

Therefore, it is not compatible with the data structure of hark_msgs/HarkWave.

Is there a practical way to make the data structure of audio_common_msgs/AudioData compatible with Hark?

Otherwise, would you suggest another package different than ros-kinetic-audio-capture in order to have Hark-compatible access to the data from the robot’s built-in microphone?

Thanks in advance.

[匿名]

Please see previous answer on how to convert Structure.

For frame proccessing, refer to the AudioStreamFromWave source code.
The source code can be obtained with the following command on Ubuntu.
apt source hark-core

Best regards,
HARK Support

[匿名]

Hello,
I’m also working on this currently.
I let audio_capture publish wave format. I tried to write a converter
from AudioData (uint8[] wave) to HarkWave following the instructions mentioned in this answer.
Here is what I ended up with:

        dhw = HarkWave()
        dhw.header.stamp = rospy.Time.now()
        dhw.header.frame_id = str(self.count)
        dhw.count = self.count
        harkwaveval = HarkWaveVal(wavedata=data_uint8)
        result = []
        result.append(harkwaveval)
        dhw.src = result 
        dhw.nch = 1 # one channel
        dhw.length = 320 
        dhw.data_bytes = 320 * 4 # float(4bytes) times length of data

When I run my network (attached below) it’s only receiving one message from ros and afterwards the network stops with the message “Normally finished”.
So my question is: Why does it stop, since there is no error regarding the network and the topic is still actively pushing messages?

Thanks in advance.

[匿名]

Hi Thomas.

Is the sheet with AUdioSreamFromRos and SaveWavePCM set to iterator?
If it is a subnet, it will only be executed once.

Best regards,
HARK support team.

[匿名]

Hi Thomas,

> When I run my network (attached below) it’s only receiving one message from ros and afterwards the network stops with the message “Normally finished”.

When creating a new sheet using HarkDesigner, by default it is set to subnet which runs the network once. Please change it to iterator so that the network runs in a loop.

If it is not too much trouble, would you mind uploading a sample of the audio you are trying to stream as well as the network file?

I noticed that you did not have a condition, are you getting any errors when running the network?

[投稿者]

投稿