/whistle-synth

zero-crossing based pitch detection for whistling

Primary LanguageShellMIT LicenseMIT

Usage

Build

  1. Check out as ~/whistle-synth

  2. Install dependencies:

    sudo apt install portaudio19-dev python3-evdev python3-mido python3-rtmidi

  3. Build it:

     make zeros-linux

It will detect pitches and generate audio.

To run on boot, /etc/systemd/system/whistle-synth.service should have:

[Unit]
Description=Pitch Detection and	Synthesis

[Service]
ExecStart=/home/jeffkaufman/whistle-synth/zeros-linux /home/jeffkaufman/whistle-synth/device-index /home/jeffkaufman/whistle-synth/current-voice /home/jeffkaufman/whistle-synth/current-volume /home/jeffkaufman/whistle-synth/current-gate
Restart=always
KillSignal=SIGQUIT
Type=simple

[Install]
WantedBy=multi-user.target

To support changing voices while headless, /etc/systemd/system/whistle-synth-kbd.service should have:

[Unit]
Description=Keyboard Control for Pitch Synthesis

[Service]
ExecStart=/usr/bin/python3 /home/jeffkaufman/whistle-synth/kbd.py
Restart=always
KillSignal=SIGQUIT
Type=simple

[Install]
WantedBy=multi-user.target

Then:

sudo systemctl enable whistle-synth-kbd
sudo systemctl enable whistle-synth
sudo systemctl daemon-reload

Set levels for consistency:

$ alsamixer
> F6 select "USB Audio Device"
> F5 [All]
> Speaker: 100
> Mic: 100
> Capture: 100

Run

  1. Put on headphones, use a directional mic, or otherwise avoid letting the output of this program mix with the input.

  2. Run it and whistle:

      make run-linux

Or

  make run-mac

It will generate audio.

Keys 0-8 on the keypad should select voices. Voices 0 through 6 expect whistling; 7 and 8 singing.

Microphone tips:

  • Works best with a directional microphone with a windscreen (vocal mics like the E835 or SM58 have one built in).

  • I use a Sennheiser E835 with an xlr to 3.5mm adapter into a USB sound card. This isn't how the microphone is designed to be used (it wants a pre-amp) but it works well enough and it's nice not to have another piece of hardware.

  • You want to be as close to the microphone as you can bear.

Raspberry PI Setup

  1. Install Raspberry Pi Os Lite (we don't want the desktop environment)
  2. sudo apt-get update && sudo apt-get upgrade
  3. sudo raspi-config
    1. "Interface Options"
      1. "Enable SSH"
    2. "Localisation Options"
      1. "WLAN Country"
    3. "System Options"
      1. "Wireless LAN"
  4. Add regular public key to ~/.ssh/authorized_keys
  5. Change default password (passwd)
  6. sudo apt install git emacs
  7. https://www.jefftk.com/p/you-should-be-logging-shell-history
  8. alsamixer
    1. select sound card "USB Audio Device"
    2. Set Speaker, Mic, and Capture to 100% volume

Future

Look into low-latency options. See http://tedfelix.com/linux/linux-midi.html and https://wiki.linuxaudio.org/wiki/raspberrypi