Raspberry based Audio Book (Pi ZeroW or 3B)

Senior (sensory impaired) adapted HW audio book player with radio and auto-generated daily update (Raspberry Pi Zero)

Motivation and requirements

My dad is practically blind and at 80 years has trouble hearing and operating tiny or more complicated electronics controls. Touch screens, smart phones, keyboards, and small mp3 players are completely out of the picture. I have tried using small dummy MP3 player (Sencor) with 5 buttons (prev, next, play|pause, volume up/down) as an initial assessment whether audio book player is something he would be able to control. Even though he uesd it, he was struggling with controlling it and the small player with 2-3x overloaded button controlls was too much. Also it lacked a fundamental option of remote book update. So I've decided to build custom player with following requirements:

  • volume control is an incremental rotary encoder
  • keep the number of buttons to minimum (spaced far apart - resilient to random touch)
  • allow remote content change - wifi
  • open content (not locked to a publisher)
  • does not need to be battery operated
  • minimal level of state indicators
  • sufficient output volume to drive speakers/headphones

Install

System

run raspi-config and set the system up
Install some system packages.

sudo apt-get install python3 evtest mpd mpc ntp vim screen git gpio

Create data dir

mkdir /data && chown pi /data && chmod 755 /data

Edit /etc/mpd.conf and change directory to /data

Set the rotary encoder overlay and disable Bluetooth and onboard audio adding to /boot/config.txt

# add rotarty encoder
dtoverlay=rotary-encoder,pin_a=19,pin_b=26,relative_axis=1

# disable builtin audio - we are using external usb card
dtparam=audio=off

# disable BT
dtoverlay=disable-bt

Crontab under user to fetch into /data/.news.mp3

crontab -e 
55 * * * * /usr/bin/curl -s -o /data/.news.mp3 http://www.adino.sk/dl/news.mp3

Flash hardening

Due to the system using flash, it might happen that the SD card eventually wears out. There are couple of things we can do to prevent this.
Disable swap and frequent writers.

apt-get remove --purge wolfram-engine triggerhappy anacron logrotate xserver-common lightdm
apt-get autoremove --purge

Edit /etc/dphys-swapfile and set CONF_SWAPSIZE=0
Verify with free, swap should be 0.

Edit /etc/mpd.conf and make log_file log to /var/lib/mpd rather than /var/log/mpd.

log_file /var/lib/mpd/mpd.log"

Modify /etc/fstab and add:

tmpfs    /tmp    tmpfs    defaults,noatime,nosuid,size=100m    0 0
tmpfs    /var/tmp    tmpfs    defaults,noatime,nosuid,size=30m    0 0
tmpfs    /var/lib/mpd    tmpfs    defaults,noatime,nosuid,size=30m    0 0
tmpfs    /var/log    tmpfs    defaults,noatime,nosuid,mode=0755,size=100m    0 0

/var/log is optional. It should not be very busy with mpd now logging to /var/lib/mpd. Use iotop -o -b -d 10 to check what is writing to the flash.

Dependencies

Use venv for managing dependencies

sudo apt-get install python3-venv
python3 -mvenv env
activate env with `source env/bin/activate`
pip3 install gpiozero
pip3 install python-mpd2
pip3 install evdev
pip3 install pyalsaaudio
pip3 install google-cloud-texttospeech

Service setup

sudo cp *.service /etc/system.d/system # or create symlinks
sudo systemctl daemon-reload
sudo systemctl enable wifi-restart
sudo systemctl enable knihaui

knihaui.py

  • User pi on Raspberry PI Zero has this repo checked out under Pi0AudioBook folder.
  • There is also folder /data on the root writable by pi user.
  • /etc/rc.local is modified to disable video output, set PCM volume to 100, set IO pins and set permissions on /data
  • We have wifi_restart.sh and related service definition to automatically ping and restart wifi.
  • /etc/systemd/system/knihaui.service takes care of running the UI.
  • Service is enabled with systemctl enable knihaui.
  • MPD is installed and enabled on the system running on port 6600 and using /data for media directory.
  • Unused or extra components are disabled. We keep avahi for name discovery.
  • To prolong SD card lifetime download overlayfs and use as per instructions in readme.

newsgen.py

  • download project certificate from google cloud to env/newsgen-credentials.json`
    To run:
  • export GOOGLE_APPLICATION_CREDENTIALS=env/newsgen-credentials.json
  • source env/bin/activate
  • Running python3 newsgen.py creates /tmp/news.mp3 if successful

Listen to Example brief in Slovak here

Automate with crontab.

V0

V0 was the set of scripts to slice larger audio books into manageable small files suitable for dumb players. This also allowed to prepend "chapter X" voice at the start of each slice.

V1

V1 is the physical build with buttons that my dad is using right now.

  • [x] Build hardware using Pi zero W
  • [x] PY UI that drives the buttons and controlls MPD
  • [x] Test remotre upgrade capability - SSH
  • [x] Add support for internet radios (SRo and Radio Litera)
  • [x] Add doc of system modification of raspbian to this doc

V2

  • [x] HW: Replace potentiometer with rotary encoder and set master volume directly using Alsa
  • [x] HW: Use bigger case - and use Raspberry PI 3B - better wifi stability
  • [x] SW: Rotary switch volume control
  • [x] SW: user request to have information about the day available as another station
  • [x] OS: read-only mount mode to prolong SD card lifetime

V3

  • [ ] HW: Replace buttons for a better quality ones
  • [ ] HW: Add 5V fan to manage temperature better
  • [ ] HW: Add serial port output to external connector for improved troubleshooting
  • [ ] HW: Add rocker switch with indicator to allow turn off/on and immediate powered-on indication
  • [ ] HW: Wire rotary selector push button for extended functionality
  • [ ] SW: Use dtoverlay for button processing - works well for rotary and simplify event processing
  • [ ] OS: Serial console

Schematic

schematic

Photos

V1

Pi0AudioBook-inside

Pi0AudioBook-top

V2

V2Top

V2Inside

GitHub

https://github.com/ahosna/Pi0AudioBook