Audio encoding

Autonomous Machines Jetson & Embedded Systems Jetson TK1
1

hi,

I am trying to record audio using a microphone connected to the audio port. First, I started by testing AAC encoding, using the following pipeline (found in the linux driver package multimedia user guide): gst-launch-0.10 audiotestsrc ! ‘audio/x-raw-int, rate=(int)44100, channels=(int)2’ ! ffenc_aac ! qtmux ! filesink location=test.mp4 -v -e . However, this does not execute as the ffenc_aac is missing. Is there a package that I need to install in order to get that encoder or should I use a different one. The audio will be saved along with mjpeg video. Thanks in advance.

2

Hi piperak,
You need to install “gstreamer0.10-ffmpeg”.

Also, audiotestsrc will generate noise. To record audio from a headset, you will need to use alsasrc.
Plus you will need to apply following amixer settings:

amixer cset name=“Stereo ADC MIXL ADC2 Switch” 0
amixer cset name=“Stereo ADC MIXR ADC2 Switch” 0
amixer cset name=“Int Mic Switch” 0
amixer cset name=“ADC Capture Switch” 1
amixer cset name=“RECMIXL BST1 Switch” 0
amixer cset name=“RECMIXR BST1 Switch” 0
amixer cset name=“RECMIXL BST2 Switch” 1
amixer cset name=“RECMIXR BST2 Switch” 1
amixer cset name=“Stereo ADC L1 Mux” “ADC”
amixer cset name=“Stereo ADC R1 Mux” “ADC”
amixer cset name=“Stereo ADC MIXL ADC1 Switch” 1
amixer cset name=“Stereo ADC MIXR ADC1 Switch” 1
amixer cset name=“Stereo ADC MIXL ADC2 Switch” 0
amixer cset name=“Stereo ADC MIXR ADC2 Switch” 0
amixer cset name=“IN1 Mode Control” “Single ended”
amixer cset name=“IN2 Mode Control” “Single ended”
amixer cset name=“Mic Jack Switch” 1

Let me know if this helps.
Regards,
Rahool

3

Hi,
I have tried to install gstreamer0.10-ffmpeg but I can’t. The package manager reports that it does not exist. I have enabled both universe and multiverse repositories. Is there another repository that i need to enable? Thanks again.

4

Looks like the Universe is the right place:

http://ports.ubuntu.com/pool/universe/g/gstreamer0.10-ffmpeg/

How did you add the Universe? I think the correct way to add Universe or Multiverse is the following:

sudo apt-add-repository universe
sudo apt-get update
sudo apt-get install gstreamer0.10-ffmpeg

The servers are different for ARM and for X86 but with apt-add-repository you don’t need to care.

5

Hi,

this is exactly what i did and the result is the following:

Package gstreamer0.10-ffmpeg is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package ‘gstreamer0.10-ffmpeg’ has no installation candidate

Is there another way to get gstreamer0.10-ffmpeg?

6

You can download with e.g. “wget” from the url I mentioned and then install with “sudo dpkg -i file.deb”. but it probably has a bunch of dependencies that you would also need to get manually if the universe and multiverse repositories are not working properly for you.

7

Is it possible to use the output of a media player (any smartphone) instead of a microphone. Do I need to alter the amixer settings in that case too and if yes are the changes similar to the ones you pointed out earlier in this post? Thanks again.

8

I assume those amixer settings enable the mic in and you can use the same settings also for recording from a smartphone. You may need to tune the recording volume but I think nothing else.

9

Hi,

I tried to use a different encoder and I was able to record sound using the following pipeline:

gst-launch -e alsasrc device=hw:1 ! audioresample ! audioconvert ! lamemp3enc target=1 bitrate=64 cbr=true ! filesink location=/home/ubuntu/audio.mp3

However, I get the following message :

WARNING: from element /GstPipeline:pipeline0/GstAlsaSrc:alsasrc0: Can’t record audio fast enough
Additional debug info:
gstbaseaudiosrc.c(840): gst_base_audio_src_create (): /GstPipeline:pipeline0/GstAlsaSrc:alsasrc0:
Dropped 8192 samples. This is most likely because downstream can’t keep up and is consuming samples too slowly.

I checked cpu utilization using top and the pipeline is reported to utilize around 10-20% of the cpu. Is this due to an alsa bug? Is there a way to resolve this. Thanks again

10

Hi piperak,

I encountered the same error, apparently it’s an issue with alsasrc. The work around is to use pulsesrc instead of alsasrc. The setup is a little different, I used “$ pact list | grep alsa_input” to get the appropriate device identifier.

I think for the built in sound card it’s:

device=alsa_input.platform-tegra-snd-rt5639.0.analog-stereo

but you should check that on your machine.

Some of the documentation that I’ve read on the Jetson states that the mic input is mono, but I think it’s really stereo.

Hope this helps

11

Hi Kangalow,

thanks for the tip.

Initially, I started with pulsesrc but sometimes I was not even able to record at all. This is the pipeline I used:
gst-launch -e pulsesrc device=“alsa_input.platform-tegra-snd-rt5639.0.analog-stereo” ! audioconvert ! lamemp3enc target=1 bitrate=64 cbr=true ! filesink location=/home/ubuntu/audio2.mp3

The device I got it by executing:
pactl list | grep -A2 ‘Source #’ | grep 'Name: ’ | cut -d" " -f2

It works most of the times. Sometimes when I use ctrl-c to break out, it refuses to stop the pipeline and the output file is 0kb. But overall it works. This is why I tried to use alsa as a more stable alternative but I guess I have to stick with pulsesrc. Thanks for the help.

12

Hi piperak,

You welcome, it sounds like you already were down the path. As far as I can figure, programming by guessing and StackOverflow is the way that most people use Gstreamer. On the Jetson there seems to be a straddle between 0.10 and 1.0, which compounds the confusion. I’ve found it very frustrating.

Cheers.