Skip to content

Recording audio in a loop

Sometimes I wish I could hear what someone just said. Most of the times, me, which is extra hard to recall with precision. We tend to remember what we meant to say, not the actual exact same words.

Normally it takes only about an hour, maybe two, to realize of the desire to rewind and listen back.

Simple audio recording

Analog mic from gaming headphones

Lets start with the basic: record from an old mic.

Lexicon beign an Intel NUC11PAHi3, it has a front 2.5 mm connector for the 4-contact headphone+mic headset, which is precisely what I have left over from the old days before I got wireless headphones.

The first question is whether the soundcard for that connector is supported and detected by the kernel:

# dmesg | egrep -i 'audio|sound|snd' | grep -vi hdmi
[    8.525267] snd_hda_intel 0000:00:1f.3: DSP detected with PCI class/subclass/prog-if info 0x040380
[    8.525325] snd_hda_intel 0000:00:1f.3: enabling device (0000 -> 0002)
[    8.527228] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
[    8.568725] snd_hda_codec_realtek hdaudioC0D0: autoconfig for ALC256: line_outs=1 (0x21/0x0/0x0/0x0/0x0) type:hp
[    8.568731] snd_hda_codec_realtek hdaudioC0D0:    speaker_outs=0 (0x0/0x0/0x0/0x0/0x0)
[    8.568732] snd_hda_codec_realtek hdaudioC0D0:    hp_outs=0 (0x0/0x0/0x0/0x0/0x0)
[    8.568734] snd_hda_codec_realtek hdaudioC0D0:    mono: mono_out=0x0
[    8.568735] snd_hda_codec_realtek hdaudioC0D0:    inputs:
[    8.568736] snd_hda_codec_realtek hdaudioC0D0:      Internal Mic=0x13
[    8.568737] snd_hda_codec_realtek hdaudioC0D0:      Internal Mic=0x12
[    8.652690] input: HDA Intel PCH Headphone as /devices/pci0000:00/0000:00:1f.3/sound/card0/input4

Note: without grep -vi hdmi there are 12 additional inputs, not the ones we want now.

/devices/pci0000:00/0000:00:1f.3/sound/card0/input4 looks like the one we want, but how do we record?

I search for a while how to do this without ALSA or Pulseadio, but in the end gave in to ALSA:

# apt install alsa-utils -y
# arecord -L | grep -C1 'HDA Intel'
hw:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Direct hardware device without any conversions
plughw:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Hardware device with all software conversions
default:CARD=PCH
    HDA Intel PCH, ALC256 Analog
    Default Audio Device
sysdefault:CARD=PCH
    HDA Intel PCH, ALC256 Analog
    Default Audio Device
front:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Front output / input
dsnoop:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Direct sample snooping device

# aplay -L | grep -C1 'HDA Intel.*Analog'
hw:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Direct hardware device without any conversions
--
plughw:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Hardware device with all software conversions
--
default:CARD=PCH
    HDA Intel PCH, ALC256 Analog
    Default Audio Device
sysdefault:CARD=PCH
    HDA Intel PCH, ALC256 Analog
    Default Audio Device
front:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    Front output / input
surround21:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    2.1 Surround output to Front and Subwoofer speakers
surround40:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    4.0 Surround output to Front and Rear speakers
surround41:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    4.1 Surround output to Front, Rear and Subwoofer speakers
surround50:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    5.0 Surround output to Front, Center and Rear speakers
surround51:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    5.1 Surround output to Front, Center, Rear and Subwoofer speakers
surround71:CARD=PCH,DEV=0
    HDA Intel PCH, ALC256 Analog
    7.1 Surround output to Front, Center, Side, Rear and Woofer speakers
--

To list the available input devices capable of capturing (recording) audio:

# arecord -l
**** List of CAPTURE Hardware Devices ****
card 0: PCH [HDA Intel PCH], device 0: ALC256 Analog [ALC256 Analog]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: S7 [SteelSeries Arctis 7], device 0: USB Audio [USB Audio]
  Subdevices: 1/1
  Subdevice #0: subdevice #0

To record from the microphone in the onboard jack:

# arecord \
  -D plughw:1,0 \
  -f S16_LE /tmp/test__.wav
Recording WAVE '/tmp/test__.wav' : Signed 16 bit Little Endian, Rate 8000 Hz, Mono
^CAborted by signal Interrupt...

To record from the microphone in the USB headphones:

# arecord \
  -D plughw:CARD=S7,DEV=1 \
  -f S16_LE /tmp/test_S7_1.wav
Little Endian, Rate 8000 Hz, Mono
^CAborted by signal Interrupt...

The problem with this is... well, these are headphone microphones, actually gaming headphones, so they only capture audio from the mouth a few milimeters from the microphone and very nearly nothing at all from anything else. This is very good performance for their intented use case, useless of today's goal.

Webcam microphone (much better)

The only other microphone available is an old webcam, never used otherwise. This begin a common Logitec C920 HD Pro, it is well supported:

[204236.991023] usb 3-2: new high-speed USB device number 4 using xhci_hcd
[204237.709488] usb 3-2: New USB device found, idVendor=046d, idProduct=0892, bcdDevice= 0.19
[204237.709496] usb 3-2: New USB device strings: Mfr=0, Product=2, SerialNumber=1
[204237.709501] usb 3-2: Product: HD Pro Webcam C920
[204237.709504] usb 3-2: SerialNumber: B9BAB6EF
[204238.583154] videodev: Linux video capture interface: v2.00
[204238.617035] gspca_main: v2.14.0 registered
[204238.621950] gspca_main: vc032x-2.14.0 probing 046d:0892
[204238.622360] gspca_vc032x: reg_r err -32
[204238.622367] vc032x: probe of 3-2:1.0 failed with error -32
[204238.622404] usbcore: registered new interface driver vc032x
[204238.634075] usb 3-2: Found UVC 1.00 device HD Pro Webcam C920 (046d:0892)
[204238.636766] input: HD Pro Webcam C920 as /devices/pci0000:00/0000:00:14.0/usb3/3-2/3-2:1.0/input/input20
[204238.637655] usbcore: registered new interface driver uvcvideo

Once again, list capture capable devices and record:

# arecord -l
**** List of CAPTURE Hardware Devices ****
card 0: PCH [HDA Intel PCH], device 0: ALC256 Analog [ALC256 Analog]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: C920 [HD Pro Webcam C920], device 0: USB Audio [USB Audio]
  Subdevices: 1/1
  Subdevice #0: subdevice #0

# arecord \
  -D plughw:CARD=C920,DEV=0 \
  -f S16_LE /tmp/test_C920.wav

At first this produce barely audible sound, need to increase the capture volume, but not too much (needed to try a few values to find a good one). More importantly, use -f cd to capture audio in a better format:

# amixer -c1 set Mic 70%
Simple mixer control 'Mic',0
  Capabilities: cvolume cvolume-joined cswitch cswitch-joined
  Capture channels: Mono
  Limits: Capture 0 - 60
  Mono: Capture 42 [70%] [41.00dB] [on]

# arecord \
  -D plughw:CARD=C920,DEV=0 \
  -f cd /tmp/test_C920.wav

Clicky noises (~10 Hz)

Audio recording devices are affected by nearby electromagnetic waves, including those produced by cables. Power cables and power extensions in particular can be most problematic, so the webcam (or microphone) must be kept well away from those.

Loop recording

This example from the arecord(1) man page shows how to record indefinitely with a new file per hour:

arecord -f cd -t wav --max-file-time 3600 --use-strftime %Y/%m/%d/listen-%H-%M-%v.wav

Record in stereo from the default audio source. Create a new file every hour. The files are placed in directories based on their start dates and have names which include their start times and file numbers.

# arecord \
  -f cd \
  -t wav \
  -D plughw:CARD=C920,DEV=0 \
  --max-file-time 3600 \
  --use-strftime %Y-%m-%d-%H-%M-%v.wav

A 5-minute recording in WAVE format takes about 50 MB, so each 1-hour file should be about 600 MB. Keeping a whole day of such files would take about 15 GB which is affordable.

Loop recording as a service

All that is left to do is keep that recording process running and making sure it starts again after each reboot.

  1. Create the script /root/bin/record-hourly to run arecord

    record-hourly
    1
2
3
4
5
6
7
8
9
    #!/bin/bash
D=/home/depot/audio/Hourly
mkdir -p $D
arecord \
-f cd \
-t wav \
-D plughw:CARD=C920,DEV=0 \
--max-file-time 3600 \
--use-strftime $D/%Y-%m-%d-%H-%M-%v.wav
    2. Create a service to run it on startup: /etc/systemd/system/record-hourly.service

    record-hourly.service
     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
    [Unit]
Description=Audio Recording
After=influxd.service
Wants=influxd.service

[Service]
ExecStart=/root/bin/record-hourly
Restart=on-failure
StandardOutput=null
User=root

[Install]
WantedBy=multi-user.target

  2. Load and start the new service

    # systemctl daemon-reload
# systemctl enable record-hourly.service
Created symlink /etc/systemd/system/multi-user.target.wants/record-hourly.service → /etc/systemd/system/record-hourly.service.
# systemctl start record-hourly.service

  3. Run a daily job to generate spectrograms from recordings, running wav-to-png as:

    wav-to-png
    1
2
3
4
5
    #!/bin/bash
for w in $*; do
  png=${w/wav/png}
  test -f $png || sox $w -n spectrogram -o $png
done

  4. Run a daily job to remove recordings older than 3 days, e.g. as clean-up-recordings

    clean-up-recordings
     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
    #!/bin/bash
D=/home/depot/audio/Hourly
test -d $D || exit 1

N=3  # Number of days to keep
now=$(date +"%s")

N_days_ago=$((now - N*24*3600))
n=1
while [[ $n -gt 0 ]]; do
  date=$(date -d @$N_days_ago +"%Y-%m-%d")
  n=$(ls $D | grep -c ${date}.*.wav) 
  rm -f $D/${date}-*.wav
  date_ts=$(date -d @$N_days_ago +"%s")
  N_days_ago=$((date_ts - 24*3600))
done