Skip to content

TODO.md

Latest commit

 

History

History
75 lines (46 loc) · 3.65 KB

TODO.md

File metadata and controls

75 lines (46 loc) · 3.65 KB

Edrumulus Road Map

The road map items are sorted by priority.

  • Investigate if a neural network (AI) can be used for Edrumulus (Github Issue 168)

    Challanges:

    • Make the network as small as possible so that it can be run on, e.g., an ESP32-S3.
    • What about latency of the algorithm? Can the AI lern to keep the delay below a certain threshold?
    • Do we have sufficient recordings or are there much more needed to lern the neural network?

  • Hot spot suppression on mesh pads with center piezo

    Striking directly on the piezo results in detected velocity values which are much too high.

    Recent recordings show that there seems to be no direct detection possibility of hot spot situations. It is more likely that we have to use the positional sensing information for this. If a position close to the middle of the pad is detected, we can apply an attenuation in case of hot spot suppression is turned on. But the positional sensing may not be precise enough to result in good detection of a hot spot strike.

  • Support direct rim strike

    See #84.

  • Introduce a first peak detection reliability

    This can be used to improve the positional sensing. E.g., if the reliability is low, we could use the position of the last detected peak if it is close to the current peak in time (e.g., if we have a fast roll situation). But the last detected peak is maybe not a good idea to use since we usually have two sticks and the position of the last peak is most probably from the other stick which position is most probably away from the correct position of the current peak.

  • Change rim shot/positinal sensing counters

    Taken from the code:

    • only use one counter instead of rim_shot_cnt and pos_sense_cnt
    • as long as counter is not finished, do check "hil_filt_new > threshold" again to see if we have a higher peak in that time window -> if yes, restart everything using the new detected peak

  • Create a super simple default drum kit in the git repo so that run_edrumulus.sh runs successfully without any other dependencies

    See #94 (comment).

  • Add support for USB OTG MIDI for the ESP32-S3

    It seems we have to wait for the Github espressif/arduino-esp32 project to support the newest ESP-IDF so that #include "tinyusb.h" does not give a compiler error. See esp-idf/examples/peripherals/usb/device/tusb_midi for an example code.

  • Improve rim shot detection

    Especially for pads like the PDA-120L where the piezos are located near the edge, the rim shot detection does perform poorly.

  • Add support for rim boost setting for multiple head sensor pads

    The rim boost setting is only supported for pads with just one head sensor.

  • Cross talk cancellation doesn't work reliably (Github Issue #109)

    User 3hhh observed that cross talk spikes may actually arrive before the spike of the pad he actually hit. -> Can be fixed with the xtalk programm created by 3hhh (see Git submodule in tools directory).

  • For the ESP32 prototype, adjust the ADC_noise_peak_velocity_scaling in edrumulus.h correctly

  • Should we consider pre-scan time high peaks for velocity estimation?

  • Documentation (user manual)

    Create an Edrumulus manual which describes the hardware/software setup, parameter description, HOWTOs, etc.

  • Algorihm documentation

    The algorithm description should be improved. Especially, improve retrigger cancellation section.